WO2020169053A1

WO2020169053A1 - Systems and methods for identifying abnormalities

Info

Publication number: WO2020169053A1
Application number: PCT/CN2020/075892
Authority: WO
Inventors: Guanqiao HE; Wei Zhang; Jialin Zhang
Original assignee: Beijing Didi Infinity Technology And Development Co., Ltd.
Priority date: 2019-02-21
Filing date: 2020-02-19
Publication date: 2020-08-27
Also published as: CN110751586A

Abstract

A system and method for identifying abnormalities. The method may include obtaining order related data. The order related data may include current order data associated with a current order and real-time state data associated with the current order (510). The method may include determining whether a current trip associated with the current order is abnormal based on the current order data and the real-time state data to generate an abnormality judgment result (520). The method may include performing a preset operation based on the abnormality judgment result (530).

Description

SYSTEMS AND METHODS FOR IDENTIFYING ABNORMALITIES

CROSS-REFERENCE TO RELATED APPLICATIONS

This application claims priority of Chinese Patent Application No. 201910130363. X, filed on February 21, 2019, the contents of which are hereby incorporated by reference.

TECHNICAL FIELD

The present disclosure generally relates to the field of on-demand transportation service, and in particular, to systems and methods for identifying abnormalities in a trip.

BACKGROUND

With the development of Internet technology, online to offline (O2O) services (e.g., online taxi services) are playing an increasingly important role in people's daily lives. With the increase of online taxi or ridesharing orders, the chance of vicious incidents that affect people’s (e.g., passengers’ or drivers’) safety or wellbeing may increase as well. Such incidents are often accompanied by abnormalities in trip resulting from the O2O service. Therefore, it is important to determine whether a trip associated with an order is abnormal as soon as possible, which may lead to the avoidance of harmful events and improve user security. It is desirable to provide methods and systems for assessing potential risks by monitoring abnormalities in a trip.

SUMMARY

According to an aspect of the present disclosure, a method for identifying abnormalities may be implemented on a computing device having one or more processors and one or more storage devices. The method may include obtaining order related data. The order related data may include current order data associated with a current order and real-time state data associated with the current order. The method may include determining whether a current trip associated with the current order is abnormal based on the current order data and the real-time state data to generate an abnormality judgment result. The method may include performing a preset operation based on the abnormality judgment result.

In some embodiments, the current order data may include at least one of identity information of a service provider, vehicle identification information related to the service provider, a service time, a starting location of the current trip, a destination of the current trip, a driving route of the current trip, or identity information of a service requester. The real-time state data may include at least one of positioning data of a terminal device associated with the current order, state data of the terminal device associated with the current order, positioning data associated with the vehicle, state data associated with the vehicle, environmental data inside the vehicle, or environmental data around the vehicle.

In some embodiments, an abnormality judgment result that the current trip associated with the current order is abnormal may include at least one of: deviation from a preset route, being in a remote area, abnormal stopping in the current trip, or abnormal driving speed.

In some embodiments, the method may include identifying an abnormality type of the current trip based on the current order data and the real-time state data. The method may include determining a danger level of the current trip based on the abnormality type of the current trip.

In some embodiments, the method may include determining a distance between a current position of the vehicle and the preset route based on positioning data associated with the vehicle and the driving route of the current trip. The method may include determining whether the vehicle deviates from the preset route based on the distance. The method may include, in response to a determination that the vehicle deviates from the preset route, determining the danger level of the current trip based on the distance, the current order data associated with the current order, and the real-time state data associated with the current order.

In some embodiments, the method may include determining a remote level of the current position of the vehicle based on the positioning data associated with the vehicle and the driving route of the current trip. The method may include determining whether the vehicle is in a remote area based on the remote level. The method may include, in response to a determination that the vehicle is in the remote area, determining a danger level of the current trip based on the remote level, the current order data associated with the current order, and the real-time state data associated with the current order.

In some embodiments, the method may include determining whether the vehicle stops abnormally in the current trip based on at least one of a count of stops in the current trip or a stopping time in the current trip. The method may include, in response to a determination that the vehicle stops abnormally in the current trip, determining a danger level of the current trip based on the count of stops in the current trip or the stopping time in the current trip, the current order data associated with the current order, and the real-time state data associated with the current order.

In some embodiments, the method may include determining whether a driving speed is abnormal based on the driving speed of the vehicle. The method may include, in response to a determination that the driving speed is abnormal, determining a danger level of the current trip based on the driving speed, the current order data associated with the current order, and the real-time state data associated with the current order.

In some embodiments, the method may include obtaining an abnormality identification model. The method may include determining the abnormality type of the current trip based on the abnormality identification model, the current order data, and the real-time state data.

In some embodiments, the method may include obtaining a plurality of first historical orders. The method may include obtaining historical order data associated with the plurality of first historical orders and historical real-time state data associated with the plurality of first historical orders. The method may include labelling at least one abnormal trip associated with the plurality of first historical orders as a positive sample, and at least one normal trip associated with the plurality of first historical orders as a negative sample. The method may include labelling an abnormality type of the positive sample. The method may include training the abnormality identification model based on the historical order data associated with the plurality of first historical orders, the historical real-time state data associated with the plurality of first historical orders, and the abnormality type of the positive sample.

In some embodiments, the method may include obtaining an abnormality evaluation model. The method may include determining the danger level of the current trip based on the abnormal evaluation model and the abnormality type of the trip.

In some embodiments, the method may include obtaining a plurality of second historical orders. The method may include obtaining historical order data associated with the plurality of second historical orders and historical real-time state data associated with the plurality of second historical orders. The method may include labelling at least one abnormal trip associated with the plurality of second historical orders as a positive sample, and at least one normal trip associated with the plurality of second historical orders as a negative sample. The method may include training the abnormality evaluation model based on the historical order data associated with the plurality of second historical orders, the historical real-time state data associated with the plurality of second historical orders, the positive sample, and the negative sample.

In some embodiments, the preset operation may include an abnormality processing operation. The abnormality processing operation may include at least one of performing an abnormality prompt to at least one of the service provider or the service requester, informing a police officer, informing an emergency contact, turning on a monitoring device in the vehicle, triggering a reporting mechanism in the terminal device, or contacting one or more service providers around the service requester for help.

According to another aspect of the present disclosure, a system for identifying abnormalities may include a data obtaining module, a risk determination module, and a risk response module. The data obtaining module may be configured to obtain order related data. The order related data may include current order data associated with a current order and real-time state data associated with the current order. The risk determination module may be configured to determine whether a current trip associated with the current order is abnormal based on the current order data and the real-time state data to generate an abnormality judgment result. The risk response module may be configured to perform a preset operation based on the abnormality judgment result.

In some embodiments, the risk determination module may further be configured to identify an abnormality type of the current trip based on the current order data and the real-time state data. The risk determination module may further be configured to determine a danger level of the current trip based on the abnormality type of the current trip.

In some embodiments, the risk determination module may further be configured to determine a distance between a current position of the vehicle and the preset route based on positioning data associated with the vehicle and the driving route of the current trip. The risk determination module may further be configured to determine whether the vehicle deviates from the preset route based on the distance. The risk determination module may further be configured to, in response to a determination that the vehicle deviates from the preset route, determine the danger level of the current trip based on the distance, the current order data associated with the current order, and the real-time state data associated with the current order.

In some embodiments, the risk determination module may further be configured to determine a remote level of the current position of the vehicle based on the positioning data associated with the vehicle and the driving route of the current trip. The risk determination module may further be configured to determine whether the vehicle is in a remote area based on the remote level. The risk determination module may further be configured to, in response to a determination that the vehicle is in the remote area, determine a danger level of the current trip based on the remote level, the current order data associated with the current order, and the real-time state data associated with the current order.

In some embodiments, the risk determination module may further be configured to determine whether the vehicle stops abnormally in the current trip based on at least one of a count of stops in the current trip or a stopping time in the current trip. The risk determination module may further be configured to, in response to a determination that the vehicle stops abnormally in the current trip, determine a danger level of the current trip based on the count of stops in the current trip or the stopping time in the current trip, the current order data associated with the current order, and the real-time state data associated with the current order.

In some embodiments, the risk determination module may further be configured to determine whether a driving speed is abnormal based on the driving speed of the vehicle. The risk determination module may further be configured to, in response to a determination that the driving speed is abnormal, determine a danger level of the current trip based on the driving speed, the current order data associated with the current order, and the real-time state data associated with the current order.

In some embodiments, the risk determination module may further be configured to obtain an abnormality identification model. The risk determination module may further be configured to determine the abnormality type of the current trip based on the abnormality identification model, the current order data, and the real-time state data.

In some embodiments, the system may include a first training module. The first training module may be configured to obtain a plurality of first historical orders. The first training module may be configured to obtain historical order data associated with the plurality of first historical orders and historical real-time state data associated with the plurality of first historical orders. The first training module may be configured to label at least one abnormal trip associated with the plurality of first historical orders as a positive sample, and at least one normal trip associated with the plurality of first historical orders as a negative sample. The first training module may be configured to label an abnormality type of the positive sample. The first training module may be configured to train the abnormality identification model based on the historical order data associated with the plurality of first historical orders, the historical real-time state data associated with the plurality of first historical orders, and the abnormality type of the positive sample.

In some embodiments, the risk determination module may further be configured to obtain an abnormality evaluation model. the risk determination module may further be configured to determine the danger level of the current trip based on the abnormal evaluation model and the abnormality type of the trip.

In some embodiments, the system may include a second training module. The second training module may be configured to obtain a plurality of second historical orders. The second training module may be configured to obtain historical order data associated with the plurality of second historical orders and historical real-time state data associated with the plurality of second historical orders. The second training module may be configured to label at least one abnormal trip associated with the plurality of second historical orders as a positive sample, and at least one normal trip associated with the plurality of second historical orders as a negative sample. The second training module may be configured to train the abnormality evaluation model based on the historical order data associated with the plurality of second historical orders, the historical real-time state data associated with the plurality of second historical orders, the positive sample, and the negative sample.

In some embodiments, the risk response module may further be configured to perform an abnormality processing operation. The abnormality processing operation may include at least one of performing an abnormality prompt to at least one of the service provider or the service requester, informing a police officer, informing an emergency contact, turning on a monitoring device in the vehicle, triggering a reporting mechanism in the terminal device, or contacting one or more service providers around the service requester for help.

According to another aspect of the present disclosure, a system may include at least one storage device storing a set of instructions and at least one processor in communication with the at least one storage device. When executing the stored set of instructions, the at least one processor may cause the system to perform a method. The method may include obtaining order related data. The order related data may include current order data associated with a current order and real-time state data associated with the current order. The method may include determining whether a current trip associated with the current order is abnormal based on the current order data and the real-time state data to generate an abnormality judgment result. The method may include performing a preset operation based on the abnormality judgment result.

According to another aspect of the present disclosure, a non-transitory computer-readable storage medium may store instructions. When executing the stored set of instructions, the at least one processor may cause the system to perform a method. The method may include obtaining order related data. The order related data may include current order data associated with a current order and real-time state data associated with the current order. The method may include determining whether a current trip associated with the current order is abnormal based on the current order data and the real-time state data to generate an abnormality judgment result. The method may include performing a preset operation based on the abnormality judgment result.

According to another aspect of the present disclosure, a device for identifying abnormalities may include at least one storage device storing a set of instructions and at least one processor in communication with the at least one storage device. When executing the stored set of instructions, the at least one processor may cause the system to perform a method. The method may include obtaining order related data. The order related data may include current order data associated with a current order and real-time state data associated with the current order. The method may include determining whether a current trip associated with the current order is abnormal based on the current order data and the real-time state data to generate an abnormality judgment result. The method may include performing a preset operation based on the abnormality judgment result.

Additional features will be set forth in part in the description which follows, and in part will become apparent to those skilled in the art upon examination of the following and the accompanying drawings or may be learned by production or operation of the examples. The features of the present disclosure may be realized and attained by practice or use of various aspects of the methodologies, instrumentalities, and combinations set forth in the detailed examples discussed below.

BRIEF DESCRIPTION OF THE DRAWINGS

The present disclosure is further described in terms of exemplary embodiments. These exemplary embodiments are described in detail with reference to the drawings. These embodiments are non-limiting exemplary embodiments, in which like reference numerals represent similar structures throughout the several views of the drawings, and wherein:

FIG. 1 is a schematic diagram illustrating an exemplary abnormality identification system according to some embodiments of the present disclosure;

FIG. 2 is a schematic diagram illustrating exemplary hardware and/or software components of a mobile device on which a terminal is implemented according to some embodiments of the present disclosure;

FIG. 3 is a block diagram illustrating an exemplary abnormality identification system according to some embodiments of the present disclosure;

FIG. 4 is a flowchart illustrating an exemplary process for preventing a risk according to some embodiments of the present disclosure;

FIG. 5 is a flowchart illustrating an exemplary process for identifying abnormalities according to some embodiments of the present disclosure;

FIG. 6 is an exemplary flowchart illustrating a process for training a machine learning model according to some embodiments of the present disclosure;

FIG. 7 is a flowchart illustrating an exemplary process for determining whether a vehicle deviates from a preset route according to some embodiments of the present disclosure;

FIG. 8 is a flowchart illustrating an exemplary process for determining whether a vehicle is in a remote area according to some embodiments of the present disclosure;

FIG. 9 is a flowchart illustrating an exemplary process for determining whether a vehicle stops abnormally in a current trip according to some embodiments of the present disclosure;

FIG. 10 is a flowchart illustrating an exemplary process for determining whether a driving speed of a vehicle is abnormal according to some embodiments of the present disclosure; and

FIG. 11 is a schematic diagram illustrating exemplary hardware and software components of a computing device according to some embodiments of the present disclosure.

DETAILED DESCRIPTION

The following description is presented to enable any person skilled in the art to make and use the present disclosure, and is provided in the context of a particular application and its requirements. Various modifications to the disclosed embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of the present disclosure. Thus, the present disclosure is not limited to the embodiments shown, but is to be accorded the widest scope consistent with the claims.

The terminology used herein is to describe particular example embodiments only and is not intended to be limiting. As used herein, the singular forms "a, " "an, " and "the" may be intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprise, " "comprises, " and/or "comprising, " "include, " "includes, " and/or "including, " when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

These and other features, and characteristics of the present disclosure, as well as the methods of operation and functions of the related elements of structure and the combination of parts and economies of manufacture, may become more apparent upon consideration of the following description with reference to the accompanying drawings, all of which form a part of this disclosure. It is to be expressly understood, however, that the drawings are for the purpose of illustration and description only and are not intended to limit the scope of the present disclosure. It is understood that the drawings are not to scale.

The flowcharts used in the present disclosure illustrate operations that systems implement according to some embodiments in the present disclosure. It is to be expressly understood, the operations of the flowchart may be implemented not in order. Conversely, the operations may be implemented in inverted order, or simultaneously. Moreover, one or more other operations may be added to the flowcharts. One or more operations may be removed from the flowcharts.

Moreover, while the system and method in the present disclosure are described primarily in regard to identifying a driving condition of a vehicle associated with a transportation service, it should also be understood that the present disclosure is not intended to be limiting. The system or method of the present disclosure may be applied to any other kind of services. For example, the system or method of the present disclosure may be applied to transportation systems of different environments including land, ocean, aerospace, or the like, or any combination thereof. The vehicle of the transportation systems may include a taxi, a private car, a hitch, a bus, a train, a bullet train, a high-speed rail, a subway, a vessel, an aircraft, a spaceship, a hot-air balloon, a driverless vehicle, or the like, or any combination thereof. The transportation system may also include any transportation system for management and/or distribution, for example, a system for sending and/or receiving an express. The application of the system or method of the present disclosure may be implemented on a user device and include a webpage, a plug-in of a browser, a client terminal, a custom system, an internal analysis system, an artificial intelligence robot, or the like, or any combination thereof. It should be understood that the application scenarios of the system and method of the present disclosure are merely some examples or embodiments of the present disclosure. For those of ordinary skill in the art, the present disclosure may also be applied to other similar scenarios based on these drawings without creative work.

The term "passenger, " "requester, " “requestor, ” "service requester, " "service requestor, " and "customer" in the present disclosure are used interchangeably to refer to an individual, an entity, or a tool that may request or order a service. Also, the term "driver, " "provider, " and "service provider" in the present disclosure are used interchangeably to refer to an individual, an entity, or a tool that may provide a service or facilitate the providing of the service. The term "user" may refer to an individual, an entity or a tool that may request a service, order a service, provide a service, or facilitate the providing of the service.

The term "service request, " "request for a service, " "requests, " and "order" in the present disclosure are used interchangeably to refer to a request that may be initiated by a passenger, a service requester, a customer, a driver, a provider, a service provider, or the like, or any combination thereof. The service request may be accepted by any one of a passenger, a service requester, a customer, a driver, a provider, or a service provider. The service request may be chargeable or free.

The term "service provider terminal, ” " terminal of a service provider, ” “provider terminal, ” and "driver terminal" in the present disclosure are used interchangeably to refer to a mobile terminal that is used by a service provider to provide a service or facilitate the providing of the service. The term "service requester terminal, " "terminal of a service requester, " “requester terminal, ” and "passenger terminal" in the present disclosure are used interchangeably to refer to a mobile terminal that is used by a service requester to request or order a service.

FIG. 1 is a schematic diagram illustrating an exemplary abnormality identification system 100 according to some embodiments of the present disclosure. The abnormality identification system 100 (also referred to as the system 100 in brevity) may be configured to determine whether a current trip associated with a current order is abnormal to generate an abnormality judgment result. The system 100 may perform a preset operation based on the abnormality judgment result to reduce the harm to a user (e.g., a driver, a passenger) associated with the current order. For example, the system 100 may be configured to determine an abnormality type of the current trip, such as whether a driving speed of a vehicle associated with the current trip is abnormal, whether the vehicle is in a remote area, etc. The system 100 may be a service platform for the Internet or other networks. For example, the system 100 may be an online service platform that provides a transportation service such as a taxi-hailing service, a chauffeur service, an express car service, a carpool service, a bus service, a driver hire service, a shuttle service, etc. In some embodiments, the system 100 may be an online service platform that provides a meal delivery service, a delivery service, a meal service, a shopping service, etc. In some embodiments, the system 100 may be an online service platform that provides a housekeeping service, a travel service, an education (e.g., offline education) service, etc. As shown in FIG. 1, the system 100 may include a processing device 110, one or more terminal (s) 120 (also referred to as a terminal device 120, a user terminal 120) , a storage device 130, a network 140, and an information source 150.

In some embodiments, the processing device 110 may process data and/or information obtained from the one or more terminal (s) 120, the storage device 130, and/or the information source 150. For example, the processing device 110 may obtain positioning information, trajectory information of the one or more terminal (s) 120, and/or feature information of a user (e.g., a driver, a passenger) related to a trip of a service order (also referred to as an order, or a current order) . The processing device 110 may process the obtained information and/or data to perform one or more functions described in the present disclosure. For example, the processing device 110 may process order related data to determine whether a trip associated with the order is in an abnormal condition. As used herein, a trip associated with the order may refer to a trip that a service requester (or a server provider) takes when the order is executed. The processing device 110 may also determine risk information based on a risk determining rule and/or an abnormality identification model to determine risk determining result and/or an abnormality judgment result. As another example, the processing device 110 may determine risk information based on a risk determining rule and/or an abnormality identification model based on the order related data. In some embodiments, the processing device 110 may determine at least one risk response operation, such as alarming and/or providing an offline support according to the risk determining result (e.g., an abnormality judgment result) .

In some embodiments, the processing device 110 may obtain order related data. The order related data may include data of a service order, for example, one or more features of the service order (also referred to as current order data associated with a current order) , real-time state data while executing the service order (also referred to as real-time state data associated with the current order, real-time state data associated with the service order) , one or more historical records associated with at least one piece of the order related data, etc. In some embodiments, the processing device 110 may process the order related data based on the risk determining rule to determine the risk determining result (e.g., an abnormality judgment result) . For example, the processing device 110 may determine whether a vehicle deviates from a preset route based on positioning data associated with the vehicle and a driving route of the current trip. As another example, the processing device 110 may whether the vehicle is in a remote area based on a remote level. In some embodiments, the processing device 110 may perform at least one risk response operation based on the risk determining result (e.g., an abnormality judgment result) .

In some embodiments, the processing device 110 may be an independent server or a server group. The server group may be centralized, or distributed (e.g., the processing device 110 may be a distributed system) . In some embodiments, the processing device 110 may be local or remote. For example, the processing device 110 may access the information and/or data stored in the terminal (s) 120, the storage device 130, and/or the information source 150 via the network 140. In some embodiments, the processing device 110 may be directly connected to the terminal 120, the storage device 130, and/or the information source 150 to access the information and/or data stored therein. In some embodiments, the processing device 110 may be implemented on a cloud platform. Merely by way of example, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an inter-cloud, a multi-cloud, or the like, or any combination thereof. In some embodiments, the processing device 110 may be implemented on the terminal (s) 120.

In some embodiments, the processing device 110 may include one or more sub-processing devices (e.g., signal-core processing engine (s) or multi-core processor (s) ) . Merely by way of example, the processing device 110 may include a central processor (CPU) , an application-specific integrated circuit (ASIC) , a special-purpose instruction processor (ASIP) , a graphics processor (GPU) , a physical processor (PPU) , a digital signal processor (DSP) , a field-programmable gate array (FPGA) , an editable logic circuit (PLD) , a controller, a microcontroller unit, a reduced instruction set computer (RISC) , a microprocessor, or the like, or any combination thereof.

In some embodiments, each of the terminal (s) 120 may be a device with data acquisition, data storage, and/or data sending functions, and may include a terminal of any user (e.g., any service requester, any service provider) , a terminal that is not directly involved in a service, a terminal of the service provider (also referred to as “a service provider terminal” ) , a terminal of the service requester (also referred to as “a service requester terminal” ) , a vehicle-mounted terminal, etc. The service provider may be an individual, a tool, or an entity that may provide a service or facilitate the providing of the service. The service requester may be an individual, a tool, or an entity that may request or order the service, or be receiving the service. For example, for a transportation service, the service provider may be a driver, a third-party platform. The service requester may be a passenger, a person other than the passenger, or a device (e.g., an Internet of Things device) that receives similar services.

In some embodiments, the terminal (s) 120 may be configured to collect various types of data, including but is not limited to service-related data (also referred to as order related data) . For example, the data collected by the terminal (s) 120 may include data related to an order (e.g., an order request time, a starting location, a destination, service requester information (e.g., passenger information) , service provider information (e.g., driver information, vehicle information, etc. ) , data related to a vehicle in a driving process (e.g., a speed of the vehicle, an acceleration of the vehicle, a posture of the vehicle, a road condition associated with the vehicle, etc. ) , data related to a trip (e.g., a preset route, an actual driving route, a cost, etc. ) , data related to service participants (e.g., service provider (s) , service requester (s) ) (e.g., personal information of the participants, information of operations performed by the service provider/service requester on the terminal (s) 120, data related to the terminal device, etc. ) , or the like, or any combination thereof. The collected data may be real-time data or historical data. The terminal (s) 120 may collect the data obtained by a sensor (e.g., a vehicle-mounted sensor) disposed on the vehicle, a sensor external to the vehicle. The terminal (s) 120 may also read data stored in its storage device or the storage device 150 via the network 140. In some embodiments, the sensor may include a positioning device, a sound sensor, an image sensor, a temperature and humidity sensor, a position sensor, a pressure sensor, a distance sensor, a speed sensor, an acceleration sensor, a gravity sensor, a displacement sensor, a torque sensor, a gyroscope, or the like, or any combination thereof.

The various types of data collected by the terminal (s) 120 may be configured to identify dangerous events and/or an abnormal condition that occur in the trip within the service time. For example, based on driving trajectory data, whether there is an abnormal stop (e.g., during the service and/or after the service) at a location, whether the signal of the user terminal is lost on a road section, whether the service is terminated in advance without arriving at the destination, whether the vehicle deviates from a preset route, whether the vehicle is in a remote area, whether the vehicle stops multiple times in the trip, whether the driving speed is slow, whether the driving time exceeds a threshold, etc., may be determined. As another example, whether the vehicle is at risk such as a collision or a rollover may be determined according to changes of the posture, the speed, and/or the acceleration of the vehicle.

In some embodiments, the terminal (s) 120 may include a tablet computer 120-1, a laptop computer 120-2, a built-in device in a vehicle 120-3, a mobile device 120-4, or the like, or any combination thereof. In some embodiments, the mobile device 120-4 may include a smart home device, a wearable device, a smart mobile device, a virtual reality device, an augmented reality device, or the like, or any combination thereof. In some embodiments, the wearable device may include a smart bracelet, a smart footgear, smart glasses, a smart helmet, a smart watch, smart clothing, a smart backpack, a smart accessory, or the like, or any combination thereof. In some embodiments, the smart mobile device may include a smartphone, a personal digital assistant (PDA) , a gaming device, a navigation device, a point of sale (POS) device, or the like, or any combination thereof. The built-in device in the vehicle 120-3 may include a vehicle-mounted computer, a vehicle data logger, a vehicle-mounted human-machine interaction (HCI) system, a driving recorder, a vehicle-mounted television, or the like, or any combination thereof.

In some embodiments, the built-in device in the vehicle 120-3 may obtain component data and/or operating data of the vehicle, such as a speed of the vehicle, an acceleration of the vehicle, a driving direction of the vehicle, a component state of the vehicle, the surrounding environment of the vehicle, or the like, or any combination thereof. The obtained data may be configured to determine whether a driving accident (e.g., a rollover, a collision) occurs, whether a vehicle is broken (e.g., an engine or a transmission part of the vehicle is broken such that the vehicle cannot move) , or the like, or any combination thereof.

In some embodiments, the terminal (s) 120 may be one or more devices with a positioning technology for positioning the location (s) of the terminal (s) 120. In some embodiments, the terminal (s) 120 may transmit the collected data/information to the processing device 110 via the network 140 for subsequent operations. In some embodiments, the terminal (s) 120 may store the collected data/information in its storage device, or transmit the collected data/information to the storage device 130 via the network 140 for storage. The terminal (s) 120 may receive and/or display notifications related to risk prevention generated by the processing device 110. In some embodiments, a plurality of terminals may be connected to each other, collect various types of data together, and preprocess the data by one or more terminals of the plurality of terminals.

The storage device 130 may store data and/or instructions. In some embodiments, the storage device 130 may store data/information obtained by the terminal (s) 120. The storage device 130 may store historical order data, such as one or more historical features of a historical service order, historical state data of a historical vehicle associated with the historical service order, a historical record associated with at least one piece of the historical order data, etc. In some embodiments, the storage device 130 may store data and/or instructions that the processing device 110 may execute or use to complete the exemplary processes described in the present disclosure. For example, the storage device 130 may store an abnormality identification model, which may determine whether the transportation service is at risk based on the data/information related to the transportation service obtained by the processing device 110. In some embodiments, the storage device 130 may store various types of order related data or historical order related data of the user terminal, for example, historical records of historical users related to historical services (e.g., such as historical evaluations) . In some embodiments, the storage device 130 may be a part of the processing device 110 or the terminal (s) 120.

In some embodiments, the storage device 130 may include a mass storage, a removable storage, a volatile read-write memory, a read-only memory (ROM) , or the like, or any combination thereof. Exemplary mass storage devices may include a magnetic disk, an optical disk, solid-state disks, etc. Exemplary removable storage may include a flash drive, a floppy disk, an optical disk, a memory card, a zip disk, a magnetic tape, etc. Exemplary volatile read-and-write memory may include a random-access memory (RAM) . Exemplary RAM may include a dynamic RAM (DRAM) , a double date rate synchronous dynamic RAM (DDR SDRAM) , a static RAM (SRAM) , a thyristor RAM (T-RAM) , and a zero-capacitor RAM (Z-RAM) , etc. Exemplary ROM may include a mask ROM (MROM) , a programmable ROM (PROM) , an erasable programmable ROM (EPROM) , an electrically erasable programmable ROM (EEPROM) , a compact disk ROM (CD-ROM) , and a digital versatile disk ROM, etc.

In some embodiments, the storage device 130 may be implemented on a cloud platform. Merely by way of example, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an inter-cloud, a multi-cloud, or the like, or any combination thereof. For example, some algorithms or data for risk determination in the present disclosure may be stored on a certain cloud platform and periodically updated. The processing device 110 may access the algorithms or the data via the network 140 to achieve the unification and interaction of the algorithms or the data of the entire platform. In some embodiments, historical data may be uniformly stored on the cloud platform to be accessed or updated by the processing devices 110 or the terminal (s) 120, thereby ensuring the data to be used in real-time by one or more platforms. For example, the terminal (s) 120 may transmit speed information and positioning information of the terminal (s) 120 to a cloud platform at any time. The system 100 may determine whether the trip associated with an order is in the abnormal condition according to feedback (s) from the terminal (s) 120.

In some embodiments, the storage device 130 may be connected to the network 140 to communicate with one or more components (e.g., the processing device 110, the terminal (s) 120, the information source 150) in the abnormality identification system 100. The one or more components in the abnormality identification system 100 may access data or instructions stored in the storage device 130 via the network 140. In some embodiments, the storage device 130 may be directly connected to or communicated with one or more components (e.g., the processing device 110, the terminal 120, the information source 150) in the abnormality identification system 100. In some embodiments, the storage device 130 may be a part of the processing device 110.

The network 140 may facilitate the exchange of information and/or data. In some embodiments, one or more components (e.g., the processing device 110, the terminal 120, the storage device 130, the information source 150) in the abnormality identification system 100 may send and/or receive information and/or data to/from other components in the abnormality identification system 100 via the network 140. In some embodiments, the processing device 110 may obtain data/information related to the transportation service from the terminal (s) 120 and/or the information source 150 via the network 140. As another example, the processing device 110 may obtain current order data associated with a current order and real-time state data associated with the current order via the network 140. As a further example, the terminal (s) 120 may obtain an abnormality identification model for determining whether the trip is at risk from the processing device 110 or the storage device 130 via the network 140. The abnormality identification model may be implemented by an application software of the terminal (s) 120. After obtaining the data/information related to the transportation service, the terminal (s) 120 may determine whether the transportation service is at risk and perform at least one risk response operation, for example, activating a telephone alarm.

In some embodiments, the network 140 may be any type of wired or wireless network, or any combination thereof. Merely by way of example, the network 140 may include a cable network, a wireline network, an optic fiber network, a telecommunications network, an intranet, an internet, a local area network (LAN) , a wide area network (WAN) , a wireless local area network (WLAN) , a metropolitan area network (MAN) , a wide area network (WAN) , a public switched telephone network (PSTN) , a Bluetooth network, a Zigbee network, a near field communication (NFC) network, a global system for mobile communications (GSM) network, a code division multiple access (CDMA) network, a time division multiple access (TDMA) network, a general packet radio service (GPRS) network, an enhanced data rate GSM evolution (EDGE) network, a wideband code division multiple access (WCDMA) networks, a high speed downlink packet access (HSDPA) network, a long term evolution (LTE) network, a user datagram protocol (UDP) network, a transmission control protocol/Internet protocol (TCP/IP) network, a short message service (SMS) network, a wireless application protocol (WAP) network, an ultra-wide band (UWB) network, a mobile communication (1G, 2G, 3G, 4G, 5G) a Wi-Fi, a Li-Fi, a narrowband Internet of Things (NB-IoT) , or the like, or any combination thereof. In some embodiments, the abnormality identification system 100 may include one or more network access points. For example, the driving condition identification system 110 may include wired or wireless network access points, via which one or more components of the abnormality identification system 100 may be connected to the network 140 to exchange data and/or information.

The information source 150 may be configured to provide an information source to the abnormality identification system 100. In some embodiments, the information source 150 may be configured to provide current order data associated with a current order and real-time state data associated with the current order, information related to the transportation service (e.g., a weather condition, transportation information, geographic information, law information and regulation information, news events, life information, life guide information, etc. ) to the abnormality identification system 100. In some embodiments, the information source 150 may be a third-party platform that may provide credit records (e.g., loan records) of the service requester and/or the service provider. The information source 150 may be implemented in a single central server, a plurality of servers or a plurality of user terminals that are connected with each other via a communication link. When the information source 150 is implemented in the plurality of user terminals, the user terminals may generate content (or referred to as "user-generated content" ) by, for example, uploading texts, voices, images, and videos to a cloud server. The information source 150 may include a plurality of user terminals and a plurality of cloud servers. The storage device 130, the processing device 110, and the terminal (s) 120 may also be used as the information source 150. For example, the terminal (s) 120 may be used as an information source to provide information on traffic conditions for other devices (e.g., the processing device 110) of the system 100.

This description of the system 100 is intended to be illustrative, and not to limit the scope of the present disclosure. Many alternatives, modifications, and variations will be apparent to those skilled in the art. The features, structures, methods, and other characteristics of the exemplary embodiments described herein may be combined in various ways to obtain additional and/or alternative exemplary embodiments. However, those variations and modifications do not depart the scope of the present disclosure.

FIG. 2 is a schematic diagram illustrating exemplary hardware and/or software components of the mobile device 200 on which the terminal (s) 120 may be implemented according to some embodiments of the present disclosure. As shown in FIG. 2, the mobile device 200 may include a communication unit 210, a display unit 220, a graphics processing unit (GPU) 230, a central processing unit (CPU) 240, an input/output 250, a memory 260, a storage 270, and a plurality of sensors 280. In some embodiments, any other suitable components, including but not limited to a system bus or a controller (not shown) , may be included in the mobile device 200.

In some embodiments, the mobile operating system 262 (e.g., IOS ^TM, Android ^TM, Windows Phone ^TM, etc. ) and one or more applications 264 may be loaded from the storage 270 into the memory 260 in order to be executed by the CPU 240. The application (s) 264 may include a browser or any other suitable mobile applications for sending data/information associated with the transportation service, and receiving and displaying information processed by or related to the abnormality identification system 100. For example, the application (s) 264 may be an online transportation platform (e.g., Didi Travel ^TM) . A user (e.g., a service requester) may request a transportation service via the application (s) 264 and send the requested information to the server. User interaction with the information stream may be achieved via the input/output 250 and provided to the processing device 110 and/or other components of the abnormality identification system 100 via the network 140.

In some embodiments, the mobile device 200 may include the plurality of sensors 280. The sensors 280 may obtain data related to service participants (e.g., drivers/passengers) , data related to the vehicle, data related to a trip associated with a service order, etc. In some embodiments, the sensors 280 may include a sound sensor, an image sensor, a temperature and humidity sensor, a position sensor, a pressure sensor, a distance sensor, a speed sensor, an acceleration sensor, a gravity sensor, a displacement sensor, a torque sensor, a gyroscope, or the like, or any combination thereof. In some embodiments, data obtained by the sensors 280 may be configured to determine whether the trip is at risk and/or determine the type of the risk. For example, the sound sensor may collect conversations between service participants. The image sensor may collect real-time scenes in the vehicle to determine whether there is a driver-passenger conflict or a property/personal safety event, for example, physical conflict, drunk driving, robbery, sexual assault, sexual harassment, etc. As another example, the position sensor may collect the real-time position of the vehicle. The displacement sensor may collect a driving trajectory of the vehicle to determine whether the trip is in an abnormal condition, such as an abnormal stop of the vehicle, deviation from a preset route, an abnormal driving time, etc. As a further example, the speed sensor, the acceleration sensor, and the gyroscope may collect a real-time speed of the vehicle, a real-time acceleration of the vehicle, a deflection amount of the terminal (s) 120, a deflection frequency of the terminal (s) 120, etc., for determining whether the vehicle is involved with an accident (e.g., a collision of the vehicle or a rollover of the vehicle) .

In some embodiments, the mobile device 200 may also communicate with (e.g., Bluetooth communication) the vehicle to obtain data (e.g., driving data of the vehicle, real-time state data) collected by the sensors disposed on the vehicle or the user terminal. The mobile device 200 may merge the data obtained from the sensor disposed on the user terminal and the data obtained from the vehicle-mounted sensor for subsequent risk determination.

In some embodiments, the mobile device 200 may send the obtained data/information, including the data obtained from the sensor (e.g., the vehicle-mounted sensor) disposed on the user terminal, to the processing device 110 of the abnormality identification system 100 for risk determination and/or risk response determination via the network 140. In some embodiments, the mobile device 200 may directly perform the risk determination and the risk response determination. For example, the application (s) 264 may include codes or modules for risk determination, and may directly perform the risk determination and the risk response determination. In some embodiments, the processing device 110 and/or the mobile device 200 of the abnormality identification system 100 may further generate a notified instruction according to the risk determining result (e.g., an abnormality judgment result) and/or the risk response determining result. The mobile device 200 may remind a user in a current status by receiving and executing the notified instruction. For example, the mobile device 200 may remind the user of a notification through voice (e.g., via a speaker) , vibration (e.g., via a vibrator) , text (e.g., via a short message service or a social application) , flashing lights (e.g., via a flash or display unit 220) , or the like, or any combination thereof.

In some embodiments, users (e.g., drivers and/or passengers) of the mobile device 200 may perform the risk determination manually. Specifically, the drivers and/or passengers may report risk (s) through the application (s) 264 in the mobile device 200. For example, a user may perform a specific operation (e.g., shaking or throwing) using the mobile device 200 to initiate an alarm procedure. As another example, an interface of application (s) 264 may include a quick entry (e.g., an alarm button, a help button) which can directly communicate with the back-end security platform. Upon determining that a user is in a dangerous situation, the user may call the police by clicking the alarm button. After the alarm, the application (s) 264 may also send the current position and trip information of the user who made the alarm to the police to assist the rescue.

To implement various modules, units, and functionalities described in the present disclosure, computer hardware platforms may be used as the hardware platform (s) for one or more of the components described herein. A computer with user interface elements may be configured to implement a personal computer (PC) or any other type of work station or terminal device, although a computer may also act as a server if appropriately programmed.

FIG. 3 is a block diagram illustrating an exemplary abnormality identification system 100 according to some embodiments of the present disclosure. As shown in FIG. 3, the system 100 may include a data obtaining module 310, a risk determination module 320, a first training module 330, a second training module 340, a risk response module 350, and an update module 360. In some embodiments, the data obtaining module 310, the risk determination module 320, the first training module 330, the second training module 340, the risk response module 350, and the update module 360 may be included in the processing device 110.

The data obtaining module 310 may obtain order related data of at least one service order. A service order may be a transportation service order (e.g., a cargo transportation order, a travel service order) that is requested, being executed, and/or completed. The order related data may include one or more features of the service order, real-time state data of a vehicle associated with the service order, a historical record associated with at least one piece of the order related data. The one or more features of the service order may include information recorded in the service order, including but being not limited to, identify information of a service provider associated with the service order, vehicle identification information associated with the service order, a service time of the service order, a starting location of a trip of the service order, a destination of a trip of the service order, a route of a trip of the service order, identify information of a service requester of the service order, an estimated cost of the service order, or the like, or any combination thereof. The real-time state data may refer to state data of a device (e.g., the vehicle) associated with the service order and/or data associated with the surrounding environment of a user inside the vehicle or the vehicle. The real-time state data may include but is not limited to location data of one or more terminals associated with the service order, state data of one or more terminals associated with the service order, state data of the vehicle, data associated with the internal environment of the vehicle, data associated with the surrounding environment of the vehicle, or the like, or any combination thereof. The historical record associated with the at least one piece of the order related data may include a historical record corresponding to a piece of data in a historical service order, for example, a historical record of a historical service provider executing a historical service order, a credit record of a historical service provider, a service requester's participation record of a historical service order, a credit record of a service requester, or the like, or any combination thereof. In some embodiments, the data obtaining module 310 may obtain the order related data by communicating with the terminal (s) 120, the storage device 130, and/or the information source 150 via the network 140. The data obtaining module 310 may transmit the order related data to the risk determination module 320 to determine a type of a risk.

In some embodiments, the data obtaining module 310 may also obtain historical order data. The historical order data may include data related to at least one historical transportation service order where a historical vehicle associated with the transportation service order was at risk. The historical order data may be the same as the order related data. The historical state data may be obtained based on the historical order data and include a type of a risk corresponding to the historical transportation service order. The type of the risk may include a robbery, a personal safety event, abnormal service cancellation, abnormal stopping during the trip, abnormal stopping after the trip, abnormal loss, abnormal delivery, abnormal trip, driving danger, or the like, or any combination thereof. In some embodiments, the historical order data may be used as training data to train an abnormality identification model or determine a risk determining rule. The trained abnormality identification model or the risk determining rule may be configured to analyze the order related data to determine whether the vehicle is at risk in a driving process of the vehicle. In some embodiments, the historical order data may be stored in the storage device 130. The data obtaining module 310 may communicate with the storage device 130 via the network 140 and obtain the historical order data stored therein.

In some embodiments, the data obtaining module 310 may extract feature information of the order related data. For example, the extraction of the feature information may refer to processing the order related data and extracting the feature information thereof. The extraction of the feature information may enhance the expression of the order related data to facilitate subsequent tasks. More description for obtaining feature information of the order related data may be found elsewhere in the present disclosure. See, e.g., FIG. 5 and descriptions thereof.

In some embodiments, the risk determination module 320 may be configured to perform a risk determination operation on a current status of the service order according to the risk determining rule. In some embodiments, the risk determining rule may be a condition/or a threshold set according to historical order data and/or user experience. The threshold of the risk determining rule may be determined according to data statistics or an intermediate result obtained during the training of an abnormality identification model. For example, in order to determine the risk of robbery and/or the risk of a female security event, a risk determination rule may be set based on a preset condition including, for example, whether an order request time is late at night, whether a starting location and a destination are in a remote area, whether a driver and/or a passenger have historical records related to the risk of robbery and/or the risk of a female security event, whether a count of sensitive words appear in the sensing data exceeds a preset value, etc.

As another example, the risk determination module 320 may determine whether the trip is in an abnormal condition, such as a collision of the vehicle, a rollover of the vehicle, based on sensor data (e.g., an acceleration of gravity) exceeding a preset threshold. In some embodiments, the risk determination module 320 may be configured to determine whether a current trip of a current service order is abnormal based on order related data of the current service order and/or state data (e.g., real-time state data) when the current service order is being executed. In some embodiments, an abnormality judgment result that the current trip associated with the current order is abnormal includes deviation from a preset route, being in a remote area, abnormal stopping in the current trip, or abnormal driving speed. In some embodiments, the phrase “in the current trip” refers to a time period starting from picking up the passenger to a current time in the trip associated with the current order. In some embodiments, the risk determination module 320 may identify an abnormality type of the current trip based on the current order data and/or state data (e.g., real-time state data) when the current order is being executed, and to determine the danger level of the current trip based on the abnormality type of the current trip.

In some embodiments, the risk determination module 320 may determine a distance between a current position of the vehicle and the preset route based on the positioning data associated with the vehicle and the driving route of the current trip. The risk determination module 320 may determine whether the vehicle deviates from the preset route based on the distance. The risk determination module 320 may determine the danger level of the current trip based on the distance, the current order data associated with the current order, and the real-time state data associated with the current order in response to a determination that the vehicle deviates from the preset route. In some embodiments, the risk determination module 320 may determine a remote level of the current position of the vehicle based on the positioning data associated with the vehicle and the driving route of the current trip. The risk determination module 320 may determine whether the vehicle is in a remote area based on the remote level. The risk determination module 320 may determine a danger level of the current trip based on the remote level, the current order data associated with the current order, and the real-time state data associated with the current order in response to a determination that the vehicle is in the remote area. In some embodiments, the risk determination module 320 may determine whether the vehicle stops abnormally in the current trip based on at least one of a count of stops in the current trip or a stopping time in the current trip. The risk determination module 320 may determine a danger level of the current trip based on the count of stops in the current trip or the stopping time in the current trip, the current order data associated with the current order, and the real-time state data associated with the current order in response to a determination that the vehicle stops abnormally in the current trip. In some embodiments, the risk determination module 320 may determine whether a driving speed is abnormal based on the driving speed of the vehicle. The risk determination module 320 may determine a danger level of the current trip based on the driving speed, the current order data associated with the current order, and the real-time state data associated with the current order in response to a determination that the driving speed is abnormal.

In some embodiments, the risk determination module 320 may use an abnormality identification model to perform the risk determination on a current state of the transportation service order. The abnormality identification model may be a machine learning model (e.g., a decision tree) . The driving abnormality identification model may be obtained after being trained based on historical order data. For example, the model may be determined by training a preliminary model based on the historical order data. For example, the model may be trained by inputting at least a part (e.g., a part, the whole) of the historical order data into the preliminary model. Actual risk information (e.g., the type of risks such as a dangerous event) of the historical order data may be obtained and used as desired output (e.g., the ground truth of the historical order data) of the preliminary model in the training process. In some embodiments, the abnormality identification model may be an integrated determination model configured to determine whether there are abnormality types including, for example, deviation from a preset route, being in a remote area, abnormal stopping in the current trip, abnormal driving speed, or the like, or any combination thereof. In some embodiments, the abnormality identification model may include a plurality of models for determining a plurality of abnormality types, respectively. For example, an abnormality identification model for an abnormality type of deviation from a preset route may be used to determine whether the vehicle associated with a current state for a transportation service deviates from the preset route. Similarly, a specific abnormality identification model may be used to determine each of other abnormality types. In some embodiments, the risk determination module 320 may determine the dangerous level of an abnormal trip using an abnormality evaluation model. The abnormality evaluation model may be a regression-based machine learning model. In some embodiments, the abnormality evaluation model may process the order related data and/or the real-time state data when the order is being executed, and output a result indicating the level of the risk, a probability of occurrence of risk, etc. The risk determination module 320 may determine one or more types of risks using a plurality of models. The plurality of models may be determined according to actual needs.

In some embodiments, the risk determining result (e.g., an abnormality judgment result) of the risk determination module 320 may include whether the vehicle being at risk and quantitative representation of the risk. Merely by way of example, the risk determining result may include at no risk, at risk, a risk probability, a type of the risk and a probability corresponding thereto, a level of the risk and a probability corresponding thereto, etc. For example, the determining result may be “at risk, deviation from a preset route-level 5” or “at risk, driving to a remote area-56%, abnormal stop-87%” . In some embodiments, the risk determination module 320 may comprehensively determine levels and/or probabilities of all risks and output a risk determining result corresponding to the comprehensive risk determination. For example, the risk determining result may be “at risk, 74%” . It should be noted that the representation of the risk determining result described above is only for illustrative purposes, and the present disclosure does not limit the representation of the risk determining result described above.

The first training module 330 may be configured to determine an abnormality identification model. In some embodiments, the first training module 330 may obtain a plurality of first historical orders. In some embodiments, the first historical orders may be obtained from one or more components of the system 100 (e.g., the storage device 130, the processing device 110, the terminal 120, the information source 150) via the network 140. In some embodiments, the first training module 330 may obtain historical order data associated with the plurality of first historical orders and historical real-time state data associated with the plurality of first historical orders. In some embodiments, historical order data associated with the plurality of first historical orders may include identity information of a historical service provider, vehicle identification information related to the historical service provider, a historical service time, a starting location of the historical trip, a destination of the historical trip, a driving route of the historical trip, identity information of a historical service requester, or the like, or any combination thereof. In some embodiments, historical real-time state data associated with the plurality of first historical orders may include positioning data of a terminal device associated with a historical order, state data of the terminal device associated with the historical order, positioning data associated with a historical vehicle, state data associated with the historical vehicle, environmental data inside the historical vehicle, environmental data around the historical vehicle, or the like, or any combination thereof.

In some embodiments, the first training module 330 may label at least one abnormal trip associated with the plurality of first historical orders as a positive sample, and at least one normal trip associated with the plurality of first historical orders as a negative sample. For example, a historical order including information about an abnormal driving event may be labeled as a positive sample. A historical order that does not include information about an abnormal driving event may be labeled as a negative sample. In some embodiments, the first training module 330 may label the positive sample and the negative sample based on a recording result. For example, the first training module 330 may label a historical order with a malignant event as a positive sample based on the recoding result. The first training module 330 may label a normal order (e.g., a historical order without a malignant event) as a negative result. In some embodiments, the positive sample may be represented as "1" and the negative sample may be represented as "0" . In some embodiments, the first training module 330 may label an abnormality type of the positive sample. The first training module 330 may train the abnormality identification model based on historical order data associated with the plurality of first historical orders, historical real-time state data associated with the plurality of first historical orders, and label abnormality type of the positive sample.

In some embodiments, the abnormality identification module may be a classification model. In some embodiments, the abnormality identification model may be a decision tree model, including but not limited to a classification and regression tree (CART) , an iterative dichotomiser 3 (ID3) , a C4.5 algorithm, a random forest, a chi-squared automatic interaction detection (CHAID) , a multivariate adaptive regression splines (MARS) , a gradient boosting machine (GBM) , or the like, or any combination thereof.

The second training module 340 may be configured to determine an abnormality evaluation model. In some embodiments, the second training module 340 may obtain a plurality of second historical orders. In some embodiments, the second training module 340 may obtain historical order data associated with the plurality of second historical orders and historical real-time state data associated with the plurality of second historical orders. In some embodiments, an output of the abnormality evaluation model may be a danger probability. In some embodiments, an output of the abnormality evaluation model may be a danger level or a danger coefficient of the current trip.

In some embodiments, the abnormality evaluation model and the abnormality identification model may be trained independently. The first training module 330 and the second training module 340 may be two independent module or an integrated module. In some embodiments, the abnormality identification model and the abnormality evaluation model may be trained together. The two models may be trained simultaneously. That is, there is no need to train the abnormality identification evaluation model and then training the abnormality evaluation model. In some embodiments, the plurality of second historical orders may be the same as or different from the plurality of first historical orders. For example, historical orders data similar with those of the abnormality identification model may be used as the training samples of the abnormality evaluation model. In some embodiments, historical orders data different from those of the abnormality identification model may be used as the training samples of the abnormality evaluation model. In some embodiments, part of the training samples of the abnormality evaluation model may be the same as the training samples of the abnormality identification model. In some embodiments, characteristic parameters of the abnormality evaluation model and characteristic parameters of the abnormality identification model may be the same. In some embodiments, the characteristic parameters of the abnormality evaluation model and the characteristic parameters of the abnormality identification model may be different. In some embodiments, part of the characteristic parameters of the abnormality evaluation model may be the same as the characteristic parameters of the abnormality identification model.

In some embodiments, the second training module 340 may label at least one abnormal trip associated with the plurality of second historical orders as a positive sample, and at least one normal trip associated with the plurality of second historical orders as a negative sample. In some embodiments, the second training module 340 may train the abnormality evaluation model based on the historical order data associated with the plurality of second historical orders, the historical real-time state data associated with the plurality of second historical orders, the positive sample, and the negative sample. In some embodiments, the abnormality evaluation model may be a logistic regression model.

In some embodiments, the risk response module 350 may include a risk ranking unit 352, a risk verification unit 354, a risk disposal unit 356, and a continuous monitoring unit 358. The risk ranking unit 352 may rank risk determining results (e.g., abnormality judgment results) based on a ranking rule. The risk ranking unit 352 may determine the ranking rule based on one or more risk parameters (e.g., a feature value such as a stop time of an abnormal stop) in different risks. The risk ranking unit 352 may also determine the ranking rule based on the risk probability and/or the level of the risk in the risk determining result. The ranking rule may also include one or more ranking thresholds (e.g., a level threshold, a probability threshold, etc. ) . The risk ranking unit 352 may rank the risk determining results according to the ranking thresholds. The risk ranking unit 352 may also determine the ranking rule based on a calculation result (e.g., a weighted mean) of a plurality of risk parameters. In some embodiments, the risk ranking unit 352 may rank the risk determining results using a ranking model. The ranking model may be a mathematical model configured to obtain a risk ranking result based on feature values of different types of the risk and/or feature values of all risks via calculation (e.g., weight calculation) . The ranking model may be a machine learning model, which may be obtained after being trained based on historical feature information associated with historical risks. In some embodiments, the risk verification unit 354 may input the risk determining results corresponding to the service orders into the trained risk ranking model to determine the ranking result. In some embodiments, the ranking result may indicate the ranking of the risk level of the service orders. In some embodiments, the ranking result may indicate the ranking of the risk probability of the service orders. In some embodiments, the ranking result may be configured to determine at least one risk response operation corresponding to the risk determining result.

In some embodiments, the risk ranking unit 352 may rank different types of the risks. For example, the risk ranking unit 352 may rank orders with the same type of the risks among all orders. The risk ranking unit 352 may obtain ranking results of different types of the risks. In some embodiments, the risk ranking unit 352 may rank all types of the risks together. For example, the risk ranking unit 352 may determine different weights for different risks. The risk ranking unit 352 may comprehensively rank the orders of the different risks according to the weights.

The risk verification unit 354 may perform risk verification. In some embodiments, the risk verification unit 354 may confirm the risks based on the ranking results of the risk ranking unit 352. For example, the risk verification unit 354 may select a preset count of orders in the front of the ranking results for the risk verification. In some embodiments, the risk verification unit 354 may directly confirm the risk based on the risk determining results of the risk determination module 320. For example, the risk verification unit 354 may perform the risk verification on orders with the risk determining result (e.g., risk level, risk probability, etc. ) of the risk determination module 320 within a preset range. In some embodiments, the risk verification unit 354 may directly perform the risk verification on all service orders.

In some embodiments, the risk verification operation may include risk verification through interaction with the user information, risk verification by the staff at the scene, obtaining audio or image information in the vehicle for the risk verification, risk verification based on traffic system broadcast information, or the like, or any combination thereof. The risk verification unit 354 may perform the risk verification manually. For an order with a potential risk, the abnormality identification system 100 may display information related to the order, and further confirm risk information of the order through a manual manner (e.g., a customer service) . In some embodiments, the risk verification unit 354 may perform the risk verification in an automated manner. For an order with a potential risk, the automatic risk verification unit 354 may confirm the risk by means of an outbound call of an interactive voice response (IVR) , a popup displayed in a terminal, an application text, a voice inquiry or voice monitoring of an in-vehicle driver and/or passenger, in-car recording and reporting, etc. In some embodiments, the risk verification unit 354 may also perform the risk verification through the manual interaction and/or automatic interaction. For an order with a potential risk, the risk verification unit 354 may perform the risk verification through telephone interaction.

In some embodiments, the risk verification unit 354 may perform a second risk verification operation. In some embodiments, the second risk verification operation may be performed when the trip is in the abnormal condition. The second risk verification may refer to verifying the abnormal trip condition, the risk information associated with the trip to obtain a verification result. If the verification result indicates that the abnormal trip condition does not exist, the abnormality judgment result may be adjusted. If the verification result indicates that the abnormal trip condition exists, the abnormality judgment result may be verified to be true. At least one abnormality processing operation of the abnormal trip condition may be continued to perform. More descriptions of risk verification operation may be found elsewhere in the present disclosure (e.g., FIG. 5 and descriptions thereof) .

The risk disposal unit 356 may perform a risk disposal operation. The risk disposal operation may include notifying emergency contacts, initiating data reporting by a service provider terminal and/or a service requester terminal, a follow-up alarm by a specialized person, performing an abnormality prompt to at least one of the service provider or the service requester, informing a police officer, informing an emergency contact, turning on a monitoring device in the vehicle, triggering a reporting mechanism in the terminal device, contacting one or more service providers around the service requester for help, or the like, or any combination thereof. In some embodiments, the risk disposal unit 356 may directly determine the risk disposal operation based on the risk determining result. For example, the risk disposal unit 356 may perform the risk disposal on high-risk orders and take different actions based on the risk probabilities of the service orders. For example, according to an algorithm, the risk disposal unit 356 may take action if the risk probability of an order exceeds 20%, for example, sending prompt information to a user terminal associated with the service order to remind a user (e.g., a driver or a passenger) that there is a risk. The risk disposal unit 356 may terminate the service if the risk probability is relatively high (e.g., higher than 90%) . In some embodiments, the risk disposal unit 356 may determine the risk disposal operation based on the risk ranking result (s) . For example, the risk disposal unit 356 may perform the risk disposal (e.g., sending a certain person to follow up) on orders with the risk ranking in the top of 30%. In some embodiments, the risk disposal unit 356 may also determine the risk disposal operation based on the risk verification result. For example, the risk disposal unit 356 may perform risk disposal operations on orders that have been identified to be at risk. The criteria and thresholds for the risk disposal may be combined with the update module 360, and dynamically adjusted based on the real-time conditions, the historical data, the feedback from the terminal (s) 120, etc.

In some embodiments, the risk disposal unit 356 may perform the risk disposal through a manner of risk research. The risk disposal unit 356 may obtain a service order and order related data of the service order that satisfies a condition for the risk research. The risk disposal unit 356 may also obtain a risk determining result of the service order and risk information of the service order. The risk disposal unit 356 may determine whether a risk event occurs in the service order based on the result of risk determination and risk information by a researcher associated with the risk research.

In some embodiments, the risk disposal unit 356 may perform the risk disposal through the manner of risk rescue. The risk disposal unit 356 may determine whether a service order satisfies a risk rescue condition based on the risk determining result. In response to determining that the service order satisfies the risk rescue condition, the risk disposal unit 356 may generate and send rescue information. For example, for an order that is determined to be at risk, the risk disposal unit 356 may obtain the risk information (e.g., risk type, risk level, etc. ) associated with the order. For an order whose risk level satisfies a level threshold, the risk disposal unit 356 may generate rescue information to notify surrounding drivers to go for helping or checking.

The continuous monitoring unit 358 may continuously monitor a service order. The continuous monitoring may be performed on a service order that is determined to be risk-free in the risk determination, one or more service orders that ranks at the end of the risk ranking results, a service order that is risk-free after risk verification, etc. In some embodiments, the continuous monitoring unit 358 may determine a terminal associated with the service order based on the order related data to be continuously monitored. The terminal may be a service provider terminal, a service requester terminal, a vehicle-mounted terminal, or the like, or any combination thereof. The continuous monitoring unit 358 may obtain text, sound, and/or image data indicating the execution of the service order through the terminal. Data may be obtained through various sensors installed in the terminal. For example, audio data may be obtained through a sound sensor (e.g., a microphone) . Video data may be obtained through an image sensor (e.g., a camera) . The obtained data may be used for the risk determination and risk disposal at a subsequent time point, for example, after 10s.

The update module 360 may update rules and/or models based on a result of the risk response operation. The updated rules may include one or more risk determining rules, one or more risk ranking rules, etc. Updated models may include an abnormality identification model, a risk ranking model, etc. In some embodiments, the update module 360 may compare the risk verification result and/or the risk disposal result with the risk determining result and/or the risk ranking result to obtain a difference. The risk parameters in the determination/ranking rule (s) may be updated according to the difference. In some embodiments, the update module 360 may determine orders in which a risk event occurs in the risk verification operation and/or the risk disposal operation as new sample data. The update module 360 may retrain the abnormality identification model based on the new sample data to update parameters thereof. At the same time, the update module 360 may retrain the risk ranking model according to feature information of each order determined based on actual ranking results obtained by the risk verification or the risk response. In some embodiments, the update module 360 may update the rules and models at a preset interval, for example, one day, one week, one month, one quarter of the year, etc. In some embodiments, the update module 360 may use an active push manner to direct the system to update.

It should be understood that the system and the modules of the system shown in FIG. 3 may be implemented in various ways. For example, the system and the modules of the system may be implemented by hardware, software, or a combination of software and hardware. As used herein, the hardware component may be implemented by dedicated logic. The software component may be stored in the storage which may be executed by a suitable instruction execution system, for example, a microprocessor or dedicated design hardware. It will be appreciated by those skilled in the art that the above processes and systems may be implemented by computer-executable instructions and/or embedded in the control codes of a processor. For example, the control codes may be provided by a medium such as a disk, a CD or a DVD-ROM, a programmable memory device such as a read-only memory (e.g., firmware) , or a data carrier such as an optical or electric signal carrier. The system in the present disclosure and the modules of the system may not only be implemented by large scale integrated circuits or gate arrays, semiconductor devices (e.g., logic chips, or transistors) , or hardware circuits of programmable hardware devices (e.g., field-programmable gate arrays, or programmable logic devices) , but may also be implemented by software executed in various types of processors, or a combination of the above hardware circuits and software (e.g., firmware) .

It should be noted that the above description of the system 100 and the modules of the system 100 is for convenience of description only, and not intended to limit the scope of the present disclosure. For persons having ordinary skills in the art, modules may be combined in various ways or connected with other modules as sub-systems, and various modifications and transformations may be conducted under the teaching of the present disclosure.

In some embodiments, the data obtaining module 310, the risk determination module 320, the first training module 330, the second training module 340, the risk response module 350, and the update module 360 disclosed in FIG. 3 may be independent modules in the system 100. In some embodiments, two or more modules may be combined as a module configured to implements the functions thereof. For example, the risk determination module 320 and the risk response module 350 may be two modules, or may be combined as a module having the functions of abnormality identification and abnormality determination. As another example, each module may share a single storage module. Each module may also have its own storage module. As a further example, the training module 330 may be omitted. However, those variations and modifications may not depart the scope of the present disclosure.

FIG. 4 is a flowchart illustrating an exemplary process for preventing a risk according to some embodiments of the present disclosure. In some embodiments, one or more operations in process 400 may be implemented in the system 100 shown in FIG. 1. For example, one or more operations in process 400 may be stored as instructions in storage device 130 and/or storage 270, and called and/or executed by the processing device 110.

In 410, the processing device 110 (e.g., the data obtaining module 310) may obtain order related data of at least one service order.

In some embodiments, a service order may be a transportation service order (e.g., a cargo transportation order, a travel service order) that is requested, being executed, and/or completed. The order related data may include one or more features of the service order (also referred to as current order data associated with a current order) , real-time state data associated with the service order (also referred to as real-time state data associated with a current order) , a historical record associated with at least one piece of the order related data, etc.

In some embodiments, the one or more features of the service order may include identify information of a service provider (e.g., a driver) , vehicle identification information related to the service order, a service time, a starting location of a trip associated with the service order, a destination of the trip associated with the service order, a driving route of the trip associated with the service order, identify information of a service requester (e.g., a passenger) , an estimated cost of the service order, or the like, or any combination thereof.

The identify information of the service provider may include the age of the service provider, a gender of the service provider, a face portrait of the service provider, contact information of the service provider, an education level of the service provider, an ID number of the service provider, a license number of the service provider, or the like, or any combination thereof. The vehicle identification information related to the service order may include a license number of the vehicle, a vehicle type, a vehicle brand, a vehicle color, a vehicle age that the vehicle has been driven, a load capacity, or the like, or any combination thereof. The service time may include a service order request time and/or a service order execution time. The service order request time may refer to a time at which the service requester issues the order request. The service order execution time may refer to a time at which the service provider starts to execute the service order. The identify information of the service requester may include the age of the service requester, a gender of the service requester, a face portrait of the service requester, contact information of the service requester, an education level of the service requester, an ID number of the service requester, positioning data of the service requester, or the like, or any combination thereof. In some embodiments, the one or more features of the service order may also include an estimated order-completed time duration, an estimated order-completed time, an estimated service cost, or the like, or any combination thereof.

In some embodiments, the order related data may include preference information of the service requester. In some embodiments, the preference information of the service requester may include a service requester’s preference for a service provider, a service requester’s preference for a departure location, a service requester’s preference for a destination, a service requester’s preference for a waiting time, or the like, or any combination thereof.

In some embodiments, the real-time state data may include positioning data of a terminal device associated with the current order, state data of the terminal device associated with the current order, positioning data associated with the vehicle, state data associated with the vehicle, environmental data inside the vehicle, environmental data around the vehicle, or the like, or any combination thereof.

In some embodiments, the positioning data of the terminal device or the vehicle may include longitudinal and latitudinal coordinates of the terminal device or the vehicle. In some embodiments, the state data associated with the vehicle may include a position of the vehicle, a driving trajectory of the vehicle, a motion state of the vehicle (e.g., whether the vehicle is stopping) , a driving speed of the vehicle, an acceleration of the vehicle, or the like, or any combination thereof. In some embodiments, the state data of the terminal device associated with the current order may include state data such as whether the service provider terminal or service requester terminal has initiated an alarm, whether the service provider terminal or service requester terminal has sent information, whether the service provider terminal or service requester terminal is normally turned on, power data of the service provider terminal or service requester terminal, a communication signal strength of the service provider terminal or service requester terminal, a sensor working status of the service provider terminal or service requester terminal, a running status of application (s) on the service provider terminal or service requester terminal, or the like, or any combination thereof.

In some embodiments, environmental data inside the vehicle may include audio data or image data (e.g., whether there is a conflict between a passenger and a driver, whether a driver is fatigued, whether a passenger is fatigued, whether a passenger is asleep, etc. ) in the vehicle. In some embodiments, environmental data around the vehicle may include data such as a real-time road condition, a traffic flow, a road type, road event information, a feature of a current location, whether a current time is late at night, or the like.

In some embodiments, the order related data may also include road data (e.g., data of roads that the vehicle is along) of the vehicle, driving behavior data of the vehicle, weather data, power data of the vehicle, or the like, or any combination thereof. For example, the road data may include a gradient of a road, a turn of a road, an attitude of a road, accident information of a road, or the like, or any combination thereof. Merely by way of example, the accident information of a road may include whether a count of accidents in the road exceeding a threshold, the type of an accident, a notification (e.g., warning information) associated with an accident, or the like, or any combination thereof. The driving behavior data may include data of a turning operation of a user driving the vehicle, data of a braking operation of a user driving the vehicle, data of using a light of a user driving the vehicle, or the like, or any combination thereof. The weather data may include rain, snow, wind, visibility, or the like, or any combination thereof. The power data of the vehicle may include power data of components of a fuel vehicle, power data of a power system of an electric vehicle, power data of a component of an electric vehicle, or the like, or any combination thereof.

In some embodiments, the real-time state data may be obtained via a sensor (e.g., a vehicle-mounted sensor) disposed on the vehicle, a terminal device (e.g., a terminal of the service provider, a terminal of the service requester) , a monitoring device external to the vehicle, or the like, or any combination thereof. For example, a driving speed of the vehicle may be detected by a wheel speed sensor disposed on the vehicle. Turning data of the vehicle may be detected by a steering wheel angle sensor disposed on the vehicle. The acceleration of the vehicle may be detected by an acceleration sensor disposed on the vehicle or a user terminal.

The real-time state data may further include operation contents of a user of a terminal (e.g., a service requester and/or a service provider) . The positioning data of the terminal device may include a position of the terminal device (e.g., the service provider terminal device, the service requester terminal device) ) related to service participants of the service order, a driving route of the vehicle, or the like.

In some embodiments, the historical record associated with the at least one piece of the order related data may include records of all service orders of the service provider, credit records of the service provider, records of all service orders of the service requester, credit records of the service requester, vehicle identification information associated with all service orders of the service provider, service times of all service orders of the service provider, starting locations of all service orders of the service provider, destinations of all service orders of the service provider, driving routes of all service orders of the service provider, costs of all service orders of the service requester, payment records of all service orders of the service requester, or the like, or any combination thereof. The records of all service orders of the service provider may include a count of completed service orders, a count of canceled service orders, a count of complaints, a count of prohibition on service providing, credit scores, evaluation levels, historical evaluation contents, or the like, or any combination thereof. The records of all service orders of the service requester may include a count of requested service orders, a count of canceled service orders, a count of completed service orders, service fee payment status, credit scores, evaluation levels, historical evaluation contents, or the like, or any combination thereof. The credit records of the service provider/service requester may include credit records of borrowing and credit card consumption.

In some embodiments, the data obtaining module 310 may obtain the order related data by communicating with the terminal (s) 120, the storage device 130, and/or the information source 150. For example, the terminal (s) 120 may acquire sensing data via various sensors installed on the vehicle and a content of the user operating on the terminal 120 in real-time. The data obtaining module 310 may perform data acquisition by communicating with the terminal (s) 120. As another example, the data obtaining module 310 may access data associated with a user (e.g., identifying information of a user) stored on the terminal (s) 120 or the storage device 130. As a further example, the data obtaining module 310 may communicate with the information source 150 to obtain data external to the terminal (s) 120.

It should be noted that the order related data may be obtained periodically (e.g., every 15 seconds, 30 seconds, etc. ) or in real-time. The data obtaining module 310 may transmit the obtained order related data to other modules (e.g., the risk determination module 320) of the processing device 110 in real-time to perform risk determination and continuous monitoring operations for different stages of service orders.

In 420, the processing device 110 (e.g., the risk determination module 320) may process the order related data and perform a risk determination operation on the at least one service order to generate a risk determining result.

In some embodiments, the risk determination may refer to determination whether a current trip associated with a current order is abnormal at a current time. An abnormality judgment result may be generated based on the risk determination operation. The abnormality judgment result that the current trip associated with the current order is abnormal may include deviation from a preset route, being in a remote area, abnormal stopping in the current trip, abnormal driving speed, danger driving, or the like, or any combination thereof.

In some embodiments, the risk determination module 320 may perform the risk determination operation on the service order based on a risk determining rule. The risk determining rule may include a condition/or a threshold set according to historical order data and/or user experience. The historical order data may include data of historical service orders whose historical trip was abnormal. Similar to the order related data, the historical order data may include a specific type of an abnormal trip of a historical service order. In some embodiments, through statistical analysis of the historical order data, the risk determining rule for the abnormal trip may be determined. For example, statistical analysis may be performed on historical order data including an event of deviation from a preset route. One or more features such as an evaluation score (e.g., a low evaluation score) of service participant (e.g., a passenger) , a distance between a position of a vehicle associated with an order and the preset route may be obtained. Then, for the determination of the event of deviation from the preset route, an evaluation score threshold and a distance threshold between a position of a vehicle associated with an order and the preset route may be determined based on the one or more features and set as the risk determining rule. In some embodiments, the thresholds of the risk determining rule may be determined according to data statistics. For example, for an order with the event of deviation from the preset route, a distance between a position of a vehicle associated with the order and the preset route basically exceeds five meters. The distance threshold between a position of a vehicle associated with an order and the preset route may be set as five meters. The risk determination module 320 may compare the order related data with the risk determining rule. The risk determination module 320 may determine a service order corresponding to the order related data with a value (e.g., a driving speed of a vehicle) exceeding the threshold as a risk order.

In some embodiments, for some types of the abnormal trips, one or more risk determining rules may be obtained. The risk determination module 320 may also determine the risk information of the vehicle by using a rule, a plurality of rules, or all rules of the one or more risk determining rules.

In some embodiments, the one or more risk determining rules may include whether a distance between a current position of the vehicle and the preset route exceeding a distance range, whether a remote level of the current position of the vehicle exceeding a level range, whether a count of stops in the current trip exceeding a count range, whether a stopping time in the current trip exceeding a time range, whether a driving speed of the vehicle exceeding a speed range, or the like. For example, in response to a determination that the distance between the current position of the vehicle and the preset route exceeds the distance range, the processing device 110 may determine that a trip is in an abnormal condition, i.e., the vehicle deviates from the preset route. As another example, in response to a determination that the remote level of the current position of the vehicle exceeds the level range, the processing device 110 may determine that a trip is in an abnormal condition, i.e., the vehicle is in a remote area. As still another example, in response to a determination that the count of stops in the current trip exceeds the count range and/or the stopping time in the current trip exceeds the time range, the processing device 110 may determine that a trip is in an abnormal condition, i.e., the vehicle stops abnormally in the current trip. As still another example, in response to a determination that the driving speed of the vehicle exceeds the speed range, the processing device 110 may determine that a trip is in an abnormal condition, i.e., the driving speed of the vehicle is abnormal.

In some embodiments, the abnormal condition and/or the risk information may be determined using a supervised learning technique, an unsupervised learning technique, a deviation analysis, etc.

In some embodiments, the risk determination module 320 may perform the risk determination on the service order to determine the abnormality type of the service order based on a machine learning model (e.g., an abnormality identification model) . The risk determination may include determining an abnormality type of the service order, a danger level of the abnormality type of the service order, an occurrence probability of the abnormality type of the service order, etc. The model may be a machine learning model, including but being not limited to, a classification and logistic regression (LR) model, a k-nearest neighbor (KNN) model, a naive Bayes (NB) model, a support vector machine (SVM) , a decision tree (DT) model, a random forest (RF) model, a classification and regression tree (CART) model, a gradient boosting decision tree (GBDT) model, a xgboost (or referred to as eXtreme Gradient Boosting) , a light gradient boosting machine (or referred to as LightGBM) , a gradient boosting machine (GBM) , a least absolute shrinkage and selection operator (LASSO) , an artificial neural network (ANN) model, etc. The model may be obtained by training a preliminary model based on historical order data (e.g., historical order data associated with the plurality of first historical orders) . For example, the model may be trained by inputting at least a part (e.g., a part, the whole) of the historical order data into the preliminary model. Actual risk information (e.g., an abnormality type such as a type of a dangerous event or abnormal condition of the historical order data may be obtained and used as desired output (e.g., the ground truth of the historical order data) of the preliminary model in the training process. Model parameters may be adjusted based on the difference between the predicted output (e.g., the predicted type of the risk) of the preliminary model and the desired output. In response to determining a condition is satisfied, the training process may be terminated. For example, the condition may include a count of training samples reaching a preset count, the prediction accuracy rate of the preliminary model exceeding a preset threshold of an accuracy rate, a value of the loss function is smaller than a preset value, or the like, or any combination thereof.

In some embodiments, the risk determination module 320 may perform the risk determination of the service order based on the abnormality identification model to determine the abnormality type of the service order. In some embodiments, the risk determination module 320 may perform the risk determination on the service order based on the abnormality identification model to determine the dangerous level of an abnormal trip of an order or the probability of the occurrence of the abnormal driving condition, etc. In some embodiments, the abnormality identification model may be a model for determining an abnormality type of a trip associated with an order. The risk determination module 320 may use the abnormality identification model to process a service order to determine whether one or more abnormality types of trips occurs in the orders. In some embodiments, an abnormality identification model configured to determine each abnormality type of a trip associated with an order may be determined. For example, a special abnormality identification model for deviation from a preset route may be used to determine whether a vehicle deviates from a preset route. Similarly, a special abnormality identification model for other abnormality types may be used to determine other abnormality types. The risk determination module 320 may use a combination of one or more special abnormality identification models to determine one or more risks. The combination of one or more special abnormality identification models may be determined based on actual needs. More detailed descriptions of the abnormality identification model may be found elsewhere in the present disclosure. See, for example, FIG. 5 and descriptions thereof.

In some embodiments, during the training process of the abnormality identification model, a generated intermediate result may be used as a determination threshold used in the risk determining rule. For example, taking a processing for training of a decision tree model configured to determine whether a vehicle is in a remote area as an example, a remote degree of a current position selected when a root node is bifurcated is used as an optimal feature for bifurcation. When the bifurcation threshold of a remote degree node in the current area reaches a stable value after corrections of multiple trainings (that is, the data of the root node may be divided into two correct categories) , this stable bifurcation threshold may be used as the determination threshold of the abnormality identification model.

In some embodiments, the risk determining result may include whether the trip being at risk, a quantitative representation of the risk, etc. The risk determining result may represent the risk information of the trip. Merely by way of example, if the trip is at risk, the risk determining result may include whether the trip being at risk, a type of the risk and a probability corresponding thereto, a level of the risk and a probability corresponding thereto, etc. For example, the determining result may be “at risk, abnormal driving speed-level 5” or “at risk, driving to a remote area-56%, abnormal stop-87%” . In some embodiments, the risk determination module 320 may comprehensively determine levels and/or probabilities of all risks and output a risk determining result corresponding to the comprehensive risk determination. For example, the risk determining result may be “at risk, 74%” . It should be noted that the representation of the risk determining result described above is only for illustrative purposes, and the present disclosure does not limit to the representation of the risk determining result described above.

In 430, the processing device 110 (e.g., the risk response module 350) may perform at least one risk response operation on the at least one service order based on the risk determining result.

In some embodiments, the risk response module 350 may perform different risk response operations according to different risk determining results. The risk response operation may include a risk ranking operation, a risk verification operation, a risk disposal operation, a continuous monitoring operation, or the like, or any combination thereof. The processing device 110 may process multiple service orders at the same time. If a large count of orders needs to be processed, the orders may be ranked to ensure higher-risk orders being processed in time. In some embodiments, the risk determining results of the service orders may be ranked. In some embodiments, one or more risk parameters may be determined based on the risk determining results. The risk determining results may be ranked based on the risk parameters. The risk parameters may include a piece of order related data (e.g., features such as a stopping time in abnormal stopping in the current trip, the longer the stopping time is, the larger a possibility that the vehicle is at risk may be) , a type of the risk, a level of the risk, a risk probability in the risk determining result, or the like, or any combination thereof.

In some embodiments, the risk ranking operation may be performed based on a ranking rule. The ranking rule may be determined based on the risk probabilities and/or the levels of the risks in the risk determining result. The ranking rule may also include ranking thresholds (e.g., a level threshold, a probability threshold, etc. ) . The risk determining results may be ranked according to the ranking thresholds, respectively. The ranking rule may be directly determined based on the risk probabilities included in the risk determining result. The ranking rule may also be determined based on a calculation result (e.g., a weighted mean) of a plurality of the risk parameters.

In some embodiments, the risk ranking operation may be performed based on a ranking model. The ranking model may be a mathematical model and configured to obtain risk ranking results based on feature values of different types of the risks and/or feature values of all risks through the calculation (e.g., weight calculation) . The ranking model may also be a machine learning model, including, but being not limited to, a classification and logistic regression (LR) model, a k-nearest neighbor (KNN) model, a naive Bayes (NB) model, a support vector machine (SVM) , a decision tree (DT) model, a random forest (RF) model, a classification and regression tree (CART) model, a gradient boosting decision tree (GBDT) model, a xgboost (or referred to as extreme Gradient Boosting) , a light gradient boosting machine (or referred to as LightGBM) , a gradient boosting machine (GBM) , a least absolute shrinkage and selection operator (LASSO) , an artificial neural network (ANN) model, etc. The model (i.e., the ranking model) may be obtained after being trained based on feature information associated with the risks. The risk response module 350 may input a plurality of risk determining results of the service orders into the ranking model to determine the ranking results. In some embodiments, the risk response module 350 may input a part or the whole of the real-time state data of trips that have been identified to be at risk to the trained ranking model to determine the ranking result.

In some embodiments, the risk response module 350 may rank different types of the risk respectively to obtain the ranking results of different types of the risks. In some embodiments, the risk response module 350 may rank all types of the risks together. For example, weights may be set for different types of the risks. The orders with different types of the risks may be ranked based on the weights to determine the risk ranking results for all service orders. In some embodiments, the risk response module 350 may rank the service orders that have been identified to be a certain type of the risks. For example, the risk response module 340 may rank service orders of which the risk determining results include robbery and personal security incidents.

In some embodiments, the risk response module 350 may directly process each service order without the risk ranking operation. The processing operation may include a risk verification operation, a risk disposal operation, a continuous monitoring operation, etc. It should be noted that the operations performed by the risk response module 350 may be different for service orders with different risk determining results. For example, for a high-risk order (e.g., the risk probability is greater than 50%) , the risk response module 350 may perform the risk disposal operation to alert the user (e.g., the service provider, the service requester) and/or directly call the police. As another example, the risk response module 350 may first perform the risk verification operation for service orders other than high-risk orders. Upon confirming that the service orders are at risk, the risk response module 340 may immediately call the police and/or rescue the user. For risk-free service orders or risk-free orders after the risk verification, the risk response module 340 may perform the continuous monitoring operation to detect whether the trip is at risk in time. In some embodiments, the risk response model 340 may perform the same operation for all orders. For example, the risk response model 340 may perform the risk verification operation for all service orders before performing other risk operations, or perform the risk disposal operation directly.

In some embodiments, the processing device 110 may perform a second risk verification operation. In some embodiments, the second risk verification operation may be performed when the trip is in the abnormal condition. The second risk verification may refer to verifying the abnormal trip condition, the risk information associated with the trip to obtain a verification result. If the verification result indicates that the abnormal trip condition does not exist, the abnormality judgment result may be adjusted. If the verification result indicates that the abnormal trip condition exists, the abnormality judgment result may be verified to be true. At least one abnormality processing operation of the abnormal trip condition may be continued to perform. In some embodiments, if the abnormal trip condition is determined based on a part of the order related data for the first time. During the second risk verification, more pieces of the order related data may be used for determining the condition of the trip associated with the current order.

In some embodiments, vehicle remote diagnosis may be used, such as calling a diagnostic program, for further verification. For example, a diagnostic program of a vehicle may be called to diagnose the abnormal trip conditions determined for the first time and determine whether the abnormal driving condition is accurate.

In some embodiments, a remote call of an APP or a vehicle control system may be configured to get more data for the risk verification. For example, the APP on the user terminal of passengers or drivers may be called to collect more data to help perform the risk verification. As another example, a vehicle control system may be called to collect more sensor data to get more data to facilitate the risk verification.

In some embodiments, APP interaction may be configured to perform the risk verification. For example, users such as passengers and drivers may use the APP to check the abnormal trip condition obtained for the first time and further perform the risk verification.

In some embodiments, automatic voice interaction may also be configured to perform the risk verification. For example, users such as passengers and drivers may perform the risk verification based on voice broadcast information associated with a feedback of the abnormal trip condition determined for the first time. For example, when the abnormal trip condition exists, a vehicle communication system or a user terminal may perform the voice broadcast. A passenger and a driver may interact with the vehicle communication system by voice, text, and feedback whether the abnormal trip condition actually exists.

In some embodiments, the purpose of the risk verification may be to determine an actual condition of the trip associated with the service order, and/or to determine whether the actual condition of the trip is consistent with the risk determining result obtained through the risk determination operation. In some embodiments, the risk verification operation may include interaction with the user for information verification, manual verification at the scene, risk verification by obtaining audio or image information in the vehicle, risk verification based on traffic system broadcast information, or the like, or any combination thereof. The user may refer to a participant of a service order, e.g., a service provider and/or a service requester. The risk verification through interaction with the user information may include risk verification through an outbound call of an interactive voice response (IVR) , a popup displayed in a terminal, application text, voice inquiries, telephone interaction, etc. For example, the outbound call of the IVR may be configured to allow a user to enter information, such as a mobile number, on a user terminal (e.g., the terminal (s) 120) to confirm that the user is in a secure state. The telephone interaction may be a call to a user to confirm a state of the user. The risk response module 350 may obtain the telephone interaction content, and determine whether a call recipient is himself or herself or whether there is a dangerous word in the telephone interaction content through techniques of voice recognition, semantic recognition, tone recognition for risk verification, etc. For example, a telephone interaction with drivers and/or passengers may be used to confirm whether the drivers or the passengers are at risk. As another example, voice information of the drivers or the passengers may be collected through anonymous calls (e.g., insurance sales, real estate sales, telephone shopping, etc. ) . The risk may be confirmed by identifying the tone (e.g., whether it is angry) , background sound, or voiceprint recognition. As another example, the telephone interaction may be conducted with a non-risk party in the vehicle (e.g., performing telephone interaction with a driver when it is determined that passengers are in danger) to confirm the risk. The risk verification by the staff to the scene may be performed based on the location of the user terminals or the vehicle of the service order. The staff near to the location may go to the scene. The audio or image information in the vehicle for risk verification may be obtained via a sensor (e.g., an image sensor, a sound sensor, etc. ) disposed on the terminal (including service provider terminal, service requester terminal, vehicle terminal, etc. ) , and then the risk verification based on the obtained audio or image information in the vehicle may be performed automatically or manually. The risk verification based on the traffic system broadcast information may be performed to confirm the authenticity of the occurrence of the risk of the service order to be confirmed through an event location, an event time and an event type in the traffic system broadcast information. In some embodiments, the risk verification operation may also include manual confirmation. The manual risk verification may be performed by displaying various information (e.g., a driving trajectory, videos and audios in the vehicle, a current position of the user, historical risk data of the user, the cause of historical risk, etc. ) of the service order to be confirmed to the back-end security confirmation personnel. The security confirmation personnel may confirm the risk information, for example, where the vehicle stopped, how many times the vehicle stopped, whether the driving trajectory disappeared, whether there were physical and/or language conflicts between users, etc.

As described above, the risk verification may confirm the actual condition of the service order to obtain a verification result. The risk verification may also confirm whether the risk determining result is consistent with the verification result. For example, in response to determining the risk determining result is consistent with the confirming result, e.g., the vehicle being not at risk, the processing device 110 may not perform any risk response operation. As another example, in response to determining the risk determining result is inconsistent with the confirming result and the confirming result is that the vehicle is not at risk, the processing device 110 may not perform any risk response operation. In response to determining the risk determining result is inconsistent with the confirming result and the confirming result is that the vehicle is at risk, the processing device 110 may perform the at least one risk response operation.

In some embodiments, the risk disposal operation may include notifying emergency contacts, initiating data reporting on driver terminal and/or passenger terminal, a follow-up alarm by a special person, performing an abnormality prompt to at least one of the service provider or the service requester, informing a police officer, informing an emergency contact, turning on a monitoring device in the vehicle, triggering a reporting mechanism in the terminal device, contacting one or more service providers around the service requester for help, or the like, or any combination thereof. Emergency contacts may be the first contact information (e.g., mobile number) of the contact when the passengers and/or drivers are in danger and added by passengers and/or drivers during registration and/or use of the service (e.g., via a passenger and/or driver terminal, a mobile app, etc. ) . For example, a quick entry for communication with the back-end security platform (e.g., an emergency contact button, an alarm button, a help button) may be set on the user terminal. If the user is determined to be in danger, the user may click the emergency contact button. After detecting that the emergency contact button is triggered, the terminal may automatically send a helping voice or text information to the emergency contact. The current positioning information of the terminal may be automatically added to the information. Alternatively, the user may alert the police by clicking the alarm button. After alerting the police, the terminal may also send the current position and trip information of the user to the police to assist the rescue. The data of the driver terminal and/or the passenger terminal may be audio, video, and image data obtained through various sensors disposed on the driver terminal and/or the passenger terminal (e.g., the terminal 120 or the mobile device 200) . The processing device 110 may obtain the data automatically. The users may also actively report this data. The follow-up alarms by a special person may be handled by a person (e.g., a manual customer service) . In some embodiments, the risk response module 350 may also perform the risk disposal operation on the service order that has undergone the risk verification. For example, if an order has been determined to be at risk, the risk response module 350 may perform the risk disposal operation such as an alarm.

In some embodiments, for a vehicle with an automatic driving function, the risk disposal operation may further include activating an automatic driving operation. The automatic driving operation may include parking on a side of a road, limiting a driving speed, controlling the vehicle to such as flashing or whistling for reminding, issuing an alarm (e.g., turning on a sound device in the vehicle) , or the like, or any combinations thereof. In some embodiments, for an electric vehicle, the risk treatment operation may further include reducing an output power of a power system of the vehicle, powering off the vehicle, or the like, or any combination thereof.

In some embodiments, the risk disposal may include risk research. The risk response module 350 may obtain a service order and order related data of the service order that satisfy a condition for the risk research. The risk response module 350 may also obtain risk determining results of the service orders and risk information related to various aspects of the service orders. The risk response module 350 may send the data to a processing device of a researcher associated with the risk research and obtain a result of the manual research through the processing device. The conditions for the risk research may include the risk determining result of the service order being that the service order is at risk, the risk level or the risk probability exceeding a research threshold, the service order without passing the risk verification, the result of the risk verification of the service order in the previous time being that the service order being not at risk (e.g. “temporary security” or “no alarm” ) but the service order is at risk in current moment, or the like. For a service order that satisfies the condition for the risk research, the risk response module 350 may obtain the risk determining result of the service order (e.g., based on operation 420) and the risk information related to various aspects of the service order, e.g., user information (e.g., a current position of the user, a count of the user being complained, etc. ) , a vehicle position (e.g., the environment of the vehicle is in a remote area, etc. ) , trajectory data (e.g., a route of the vehicle deviates from a common route, the vehicle stops in a location for too long, etc. ) , information extracted inside the vehicle (e.g., recording, video, call, image, etc. ) , external correlation information of the vehicle (e.g., a traffic flow, etc. ) . After obtaining the information, the risk response module 350 may send the data to the processing device of the researcher associated with the risk research. After receiving the data, the processing device of the researcher may automatically research the service order to determine whether a dangerous event and/or an abnormal condition occurs, or the researcher may determine the result by operating the processing device. In some embodiments, the risk response module 350 may generate a risk-research order and assign the risk-research order to a plurality of processing devices of the researcher to perform the risk research to determine the result of the risk research. The risk-research order may be displayed in a preset form (e.g., a list) in an interface (e.g., in a processing interface of a processing device of a researcher) . A back-end security researcher may select or click the list to view the information contained in the risk-research order. For example, the risk determining result of the service order of the risk-research order and the risk information related to various aspects of the service order may be generated. Whether a dangerous event and/or an abnormal condition occurs may be determined. At the same time, the information may be in the form of highlighting, for example, changes in the color, the thickness of the font, etc. In some embodiments, the risk response module 340 may first determine the service order that satisfies the condition for the risk research, and send the determining result in the form of a system opinion together with the risk-research order to the processing device of the researcher to assist the determination.

In some embodiments, the risk disposal may also include risk rescue. The risk response module 350 may generate rescue information based on relevant information and a risk determining result of a service order at risk and to be disposed. In some embodiments, the risk response module 350 may determine whether the service order satisfies a risk rescue condition based on the risk determining result. The risk response module 350 may determine that a service order with a risk level and/or a risk probability exceeding a rescue threshold (for example, 80%, 85%, or 90%) satisfies the risk rescue condition. For a service order that satisfies the rescue conditions, the risk response module 350 may generate rescue information based on the relevant information of the service order. For example, the risk response module 350 may generate rescue information based on the vehicle position, vehicle information, the type of risk that occurred in the service order, etc. For example, a white vehicle whose current position is near the east gate of Central Park and whose license plate number is Beijing A12345 may be in an abnormal stopping, suspected of robbery, please go to check and rescue. After generating the rescue information, the risk response module 350 may send the rescue information to a processing device associated with the police, a terminal associated with an emergency contact, a terminal associated with another service provider, etc. When the processing device associated with the police sends the rescue information, the police may be alerted at the same time. When sending the rescue information to the terminal associated with the emergency contact, reminder information may be sent at the same time to remind the emergency contact to report to the police or to ensure personal safety during checking and/or rescuing. The other service providers may include service providers whose location does not exceed a first distance threshold from the current execution location of the service order at risk and to be disposed. The current execution location may refer to the location of a participant (including users and vehicles) of the service order at risk and to be disposed at the current moment. In some embodiments, in addition to sending the rescue information, subsidy information or reward information may also be sent for reminding the service providers (e.g., drivers) that they may receive a subsidy or reward if they go to check and/or rescue. In some embodiments, different counts and different types of drivers may be notified for different risk events. For example, a count of drivers notified to rescue due to an abnormal stop event may be far smaller than a count of drivers notified to rescue due to a robbery event. At the same time, the drivers being sent to check and rescue a robbery event may be young drivers. In some embodiments, the rescue information may be sent in consideration of the distance of other drivers from the location where the risk event occurred and conditions along a route including the location.

In some embodiments, the risk response operation may be delayed. By collecting the security behavior of the user within the delay time, the pressure and impact on a risk processing device (e.g., processing device 110) may be reduced. Since the processing device 110 may need to process a plurality of service orders at the same time, the delay processing may reduce the load of the processing device 110 and speed up the processing speed of the service orders. In some embodiments, after a service order that is determined to be at risk is completed, the risk response module 340 may obtain data indicating user behaviors of a user associated with the service order. The risk response module 350 may also determine whether the user associated with the service order has performed a security behavior based on data indicating the user behaviors associated with the service order. If the security behavior is performed by the user associated with the service order, a risk determining result that the service order is at risk may be canceled. For example, in operation 420, the service order may be determined to be with the risk of an abnormal stop. The risk level of the abnormal stopping may be a mild level (e.g., the risk level and the risk probability of the abnormal stopping within a preset threshold range) , the risk response module 340 may continue to monitor the service order. If a driver associated with the service order is able to continue to accept orders normally and/or a passenger associated with the service order is able to continue to request an order normally after the current order is completed, the risk determining result that the service order is at the risk of the abnormal stop risk may be canceled. The driver and/or the passenger may be determined to be safe. In some embodiments, the order determined to be at high risk may also be verified during the delay time. For example, the verification may be performed through manners such as manual verification, automatic verification, phone-based interactive verification, etc. For example, the verification may include guiding the passenger to confirm whether there is a security risk on the passenger terminal (e.g., send information to be answered in the APP, initiate a red envelope grab activity, etc. ) , dialing a service phone number automatically, making an indirect call (e.g., to obtain relevant information by calling a financial service phone, etc. ) , contacting a relative or a friend to verify.

In some embodiments, the user may independently determine and report the security risk. For example, the interface of the application 380 may include a quick entry (e.g., an alarm button, a help button) that communicates directly with the online-to-offline service platform. The user may report risks through the quick entry. As another example, the user may perform a specific operation (e.g., pressing, shaking, or throwing) on the mobile device 200. When the sensors (e.g., sound sensors, image sensors, pressure sensors, speed sensors, acceleration sensors, gravity sensors, displacement sensors, gyroscopes, or the like, or any combination thereof) disposed in the mobile device 200 identify the specific operation, the mobile device 200 may start an alarm procedure to report the security risk. After receiving the report, the risk response module 350 may determine the accuracy of the reported security risk (e.g., whether there is noise, etc. ) for the risk verification and the risk disposal.

In some embodiments, the risk disposal may also include continuous monitoring. The continuous monitoring may be performed on a service order that is determined to be risk-free in the operation 420, a part of service orders that ranks at the end of the risk ranking results, a service order that is risk-free after the risk verification, etc. In some embodiments, the risk response module 350 may determine a terminal associated with the service order based on the order related data to be continuously monitored. The terminal may be a service provider terminal associated with the service order, a service requester terminal associated with the service order, a vehicle terminal, or the like. The risk response module 350 may obtain text, sound, and/or image data indicating the execution of the service order through the terminal. Data may be obtained through various sensors installed on the terminal. For example, audio data may be obtained through a sound sensor (e.g., a microphone) . Video data may be obtained through an image sensor (e.g., a camera) . The obtained data may be used for the risk determination and risk disposal at a subsequent time point, for example, after 10s.

It should be noted that the risk determination and risk response to an order is an ongoing process. When a service order is determined to be safe at the current moment or during the risk response operation (e.g., a risk verification operation) , the continuous monitoring may still be performed. The risk determination and the risk response may be repeated to determine whether a subsequent risk event occurs. For example, the risk determination and subsequent operations (e.g., the risk verification operation, the risk response operation) may be performed at every preset time (e.g., 10 seconds) . When the service order is completed for a threshold time (e.g., 10 minutes, 20 minutes, 30 minutes) , the risk determination and the risk response operation for the order may be stopped. At the same time, the risk response module 350 may continuously monitor the service order with risk determining result being risk-free obtained in operation 420.

Similarly, it should be understood that the processing operations in the risk response may be selectively performed. In some embodiments, the risk response module 350 may rank all service orders based on the risk determining results, and then selectively perform subsequent operations according to the ranking results. For example, the risk response module 350 may perform risk disposal operations on service orders that rank in front of the ranking result. The risk response module 350 may perform risk disposal operations on service orders that rank in the middle of the ranking result. The risk response module 350 may perform continuous monitoring operations on service order that ranks at the end of the ranking result. In some embodiments, the risk response module 350 may skip the ranking operation, directly perform risk verification on all service orders, and perform subsequent processing operations based on the verification results. For example, a risk-free service order may be continuously monitored after the risk of the risk-free service order being confirmed. For a risk service order, a user associated with the risk service order may be selectively reminded the user (e.g., the risk including an abnormal stop of the vehicle) or an alarm may be directly initiated (e.g., the risk of robbery) according to the risk level of the risk service order. In some embodiments, the risk response module 350 may directly process all service orders based on the risk determining result. For example, the risk response module 350 may send an alert to related users of service orders with risk determining results of lower risks. For service orders with risk determining results of higher risks, the risk response module 350 may directly call the police. For risk-free service orders, the risk response module 350 may perform continuous monitoring to discover subsequent risks in the shortest time when the subsequent risks occur. In some embodiments, the risk response module 350 may rank the service orders based on the risk determining results, and directly process the service orders based on the ranking results. For example, the risk response module 350 may first process the service orders in the front of the ranking results (e.g., orders with higher risks) , and then continue to process the order at the end of the ranking result (e.g., orders with lower risks) after the processing of the service orders with the higher risks is completed. In some embodiments, the risk response module 350 may perform delay processing on the service order based on the risk determining result. For example, the risk response module 350 may monitor the service order that is determined to be at risk. After the service order is completed, the risk response module 350 may obtain the behavior data of the user related to the service order. If the user has a security behavior, for example, a user related to the service order continues to request a transportation service after the service order is completed, the risk response module 350 may determine that the service order is a secure order.

In 440, the processing device 110 (e.g., the update module 360) may update rules and/or models based on the risk response operation.

In some embodiments, the rules may include one or more risk determination rules, one or more risk ranking rules, or the like. The models may include an abnormality identification model, a risk ranking model, or the like. In some embodiments, the update module 360 may compare the risk verification result and/or the risk disposal result with the risk determining result to obtain differences. The risk parameter in the risk determination rule may be updated according to the differences. For example, the risk determination rule for determining a robbery event may be determined based on a service request time of a service order associated with the robbery event and a starting location of the service order associated with the robbery event. If the service request time is after 12 p. m. and a destination of the service order is in a nearby city, the risk determining result may be that a trip associated with the service order is at the risk of robbery. When the risk of robbery is determined by performing risk verification on a plurality of service orders and no event of robbery is found in service orders with service request times between 12 pm and 12: 30 pm. Then the update module 360 may change the risk determining rule that a service order with a service request time later than 12: 30 in the evening and a destination of the service order located in a nearby city or country may have the risk of robbery. In some embodiments, the update module 360 may determine the orders in which the risk event occurs in the risk verification operation and/or the risk disposal operation as new sample data to retrain the abnormality identification model to update the parameters of the model. Similarly, for the risk ranking rules and the training of the risk ranking model, the update module 360 may also compare the risk verification result and/or the risk disposal result with the risk ranking results to obtain differences and update the ranking rules and/or the model. For example, if a high-risk order in the front of the ranking result is determined to have no risk in subsequent risk verification operations, the update module 360 may update the risk parameters used for the ranking. For updating the risk ranking model, the update module 360 may retrain the risk ranking model according to feature information of each order with an actual ranking result obtained through the risk verification or risk response operations to achieve the purpose of updating. In some embodiments, the rules and models may be updated at preset intervals, for example, one day, one week, one month, one quarter, or the like.

It should be noted that the above description is merely provided for the purposes of illustration, and not intended to limit the scope of the present disclosure. For persons having ordinary skills in the art, multiple variations and modifications may be made under the teachings of the present disclosure. However, those variations and modifications do not depart from the scope of the present disclosure. In some embodiments, one or more other optional operations may be omitted in the process 400. For example, for service orders with risk determining results being at high risk (e.g., risk levels risk probabilities of the service orders, etc. are higher than a preset threshold) , the risk ranking operation and the risk verification operation may be omitted, and the risk disposal operation may be directly performed (e.g., call the police or refer to a security officer) . As another example, for service orders with risk determining results being at low risk (e.g., risk levels, risk probabilities, etc. are lower than the preset threshold) , the monitoring and delay processing may be performed (e.g., continue to perform the data acquisition and execute the risk determination again after a preset time) . As still another example, operation 440 may be omitted.

FIG. 5 is a flowchart illustrating an exemplary process for identifying abnormalities according to some embodiments of the present disclosure. In some embodiments, one or more operations in process 500 may be implemented in the system 100 shown in FIG. 1. For example, one or more operations in process 500 may be stored in the form of instructions in the storage device 130 and/or storage 270 and executed by processing device 110.

In 510, the processing device 110 (e.g., the data obtaining module 310) may obtain order related data. In some embodiments, the order related data may at least include current order data associated with a current order and/or real-time state data associated with the current order. In some embodiments, the current order data may include identity information of a service provider, vehicle identification information related to the service provider, a service time, a starting location of a current trip, a destination of the current trip, a driving route of the current trip, identity information of a service requester, or the like, or any combination thereof. In some embodiments, the real-time state data of current order may include positioning data of a terminal device associated with the current order, state data of the terminal device associated with the current order, positioning data associated with the vehicle, state data associated with the vehicle, environmental data inside the vehicle, environmental data around the vehicle, or the like, or any combination thereof. In some embodiments, the terminal device associated with the current order may be a service provider terminal, such as a mobile terminal of a driver. In some embodiments, the terminal device associated with the current order may be a service requester terminal, such as a mobile terminal of a passenger. More descriptions for obtaining order related data may be found elsewhere in the present disclosure (e.g., FIGs. 3 and 4, and descriptions thereof) .

In some embodiments, the processing device 110 may extract feature information of the order related data. For example, the extraction of the feature information may refer to processing the order related data and extracting the feature information thereof. The extraction of the feature information may enhance the expression of the order related data to facilitate subsequent tasks.

In some embodiments, the processing device 110 may extract the feature information based on a feature extraction algorithm. For example, the feature extraction algorithm may include a statistical algorithm (e.g., a principal component analysis algorithm) , a dimension reduction algorithm (e.g., a linear discriminant analysis algorithm) , etc.

In some embodiments, the feature information of the order related data may include feature information of a driving road of the vehicle, feature information of a driving behavior associated with the vehicle, feature information of weather in a driving region associated with the vehicle, feature information of the power of the vehicle, feature information of the location of the vehicle, feature information of a state of a user terminal associated with the vehicle, feature information associated with the environmental data inside the vehicle, feature information associated with the environmental data around the vehicle, feature information of the service provider, feature information of the service requester, feature information of a service order associated with the vehicle, or the like, or any combination thereof.

In some embodiments, the feature information may include one or more numeric values, one or more vectors, one or more determinants, one or more matrices or the like, or any combination thereof. For example, a driving age of a service provider may be converted into a feature value or a feature vector of a proficiency level that the service provider drives a vehicle.

In some embodiments, the order related data may be converted into the feature values according to one or more rules. Take the driving age of the service provider as an example, if the driving age is within 0-3 years, a feature value of the proficiency level may be within a range of 0 and 0.6. If the driving age is within 3-6 years, a feature value of the proficiency level may be within a range of 0.6 and 1. If the driving age is within 3-6 years, a feature value of the proficiency level may exceed 1.

In some embodiments, the order related data may be converted into the feature information according to a continuous function. Take the driving age of the service provider as an example, a sigmoid function may be used as a feature value of the driving age.

In some embodiments, a bucketing manner may be configured to convert the order related data into one or more feature vectors. Taking the driving age of the service provider as an example, a driving year of 0-3 years may proportionally correspond to [1, 0, 0] . A driving year of 3-6 years may proportionally correspond to [0, 1, 0] . A driving year of more than 6 years may proportionally correspond to [0, 0, 1] .

In some embodiments, the feature information may be determined using multi-source data. For example, a feature value or a vector representing the influence of wind force may be obtained based on the wind direction, the wind force, and the driving state of the vehicle. The influence value when an angle between the wind direction and the direction of the vehicle is 90 degrees may be greater than when the wind direction and the direction of the vehicle are the same. As another example, the greater the vehicle speed is, the greater the influence of the wind force may be.

In some embodiments, the feature information may be extracted using a combination of multiple vectors. The combination of the vector may refer to combining feature values or feature vectors having related relationships into a new feature value or a feature vector, which may represent the feature information better. For example, a feature value representing traffic flow, a feature value representing road type, and a feature value representing road event information, may be weighted and summed to obtain a combed feature value representing road condition.

In some embodiments, representation learning may also be performed on the order related data to extract the feature information. Representing learning may refer to that a model automatically learns input data (e.g., the real-time status data) to obtain features, which may facilitate the extraction of the feature information.

In some embodiments, at least a part (e.g., a part, the whole) of the feature information may be obtained based on historical data or historical features using a machine learning model. For example, a value representing an abnormal speed of a service provider may be obtained based on a feature of a driving action of the driver, a feature of climate effect, a feature of a time point of the service provider driving the vehicle, a feature of a vehicle condition, etc., using a driving stability model. In some embodiments, the machine learning model may be a regression model determined based on linear regression and neural networks. In some embodiments, the machine learning model may determine the feature information by convolution or pooling.

In 520, the process device 110 (e.g., the risk determination module 310) may determine whether a current trip associated with the current order is abnormal based on the current order data and/or the real-time state data to generate an abnormality judgment result. In some embodiments, the abnormality judgment result that the current trip associated with the current order is abnormal may include deviation from a preset route, being in a remote area, abnormal stopping in the current trip, abnormal driving speed, or the like, or any combination thereof.

In some embodiments, the processing device 110 may determine whether a vehicle deviates from the preset route. In some embodiments, the deviation from the preset route may refer to a distance between a current position of the vehicle and the preset route. In some embodiments, when the distance between the current position of the vehicle and the preset route exceeds a distance threshold, the processing device 110 may determine that the vehicle deviates from the preset route. In some embodiments, when a duration that the vehicle deviates from the preset route exceeds a duration threshold, the processing device 110 may determine that the vehicle deviates from the preset route. For example, when the duration that the vehicle deviates from the preset route exceeds the duration threshold (e.g., 10 minutes) , the processing device 110 may determine that the vehicle deviates from the preset route.

In some embodiments, the processing device 110 may determine whether the vehicle is in the remote area. In some embodiments, the processing device may determine whether the vehicle is in the remote area based on a remote level of a current position of the vehicle. For example, when the remote level of the current position of the vehicle exceeds a preset degree, the processing device 110 may determine that the vehicle is in the remote area. As another example, when a remote level of at least one position of positions related to a driving route exceeds the preset degree, the processing device 110 may determine that the vehicle is in the remote area.

In some embodiments, the processing device 110 may determine whether the vehicle stops abnormally in the current trip. In some embodiments, when a count of stops of the vehicle within a period exceeds a preset number, the processing device 110 may determine that the vehicle stops abnormally in the current trip. In some embodiments, when a stopping time in a single stop in the current trip exceeds a single stopping threshold, the processing device may determine that the vehicle stops abnormally in the current trip. In some embodiments, when a total stopping time in the current trip exceeds a total stopping threshold, the processing device 110 may determine that the vehicle stops abnormally in the current trip. In some embodiments, when a distance between two positions of the vehicle where the vehicle stops successively is less than a distance threshold, the processing device 110 may determine that the vehicle stops abnormally in the current trip. In some embodiments, the processing device 110 may determine whether the vehicle stops abnormally in the current trip based on a stopping position of the vehicle. For example, when the vehicle stops at a position deviating from the preset route for a relatively long time (e.g., a stopping time exceeds a stopping time threshold) , the processing device 110 may determine that the vehicle stops abnormally in the current trip. As another example, when the vehicle stops at a position with a high remote level for a relatively long time (e.g., a stopping time exceeding a stopping time threshold) , the processing device 110 may determine that the vehicle stops abnormally in the current trip.

In some embodiments, the processing device may determine whether a driving speed of the vehicle is abnormal. In some embodiments, when an average driving speed of the vehicle within a time period exceeds a first speed threshold, the processing device 110 may determine that the driving speed is abnormal. For example, if the average driving speed of the vehicle within 15 minutes exceeds 130 km/h and the vehicle is in an overspeed driving state for a relatively long time (e.g., a driving time exceeds a time threshold) , the processing device 110 may determine that the driving speed is abnormal. In some embodiments, when the average driving speed of the vehicle within time period is less than a second speed threshold, the processing device 110 may determine the driving speed is abnormal. For example, if the average driving speed of the vehicle within 15 minutes is less than 40 km/h, and the vehicle is in a low-speed driving state for a relatively long time (e.g., a driving time exceeds a time threshold) , the processing device 110 may determine that the driving speed is abnormal. In some embodiments, when an actual driving speed of the vehicle exceeds a speed threshold for a time period, the processing device 110 may determine that the driving speed is abnormal. For example, if the actual speed of a vehicle exceeds 120 km/h for 10 minutes, the processing device 110 may determine the driving speed is abnormal. In some embodiments, when the actual driving speed of a vehicle is below the speed threshold for a time period, the processing device 110 may determine the driving speed is abnormal. For example, if a driving speed of a vehicle is less than 35 km/h for 10 minutes, the processing device 110 may determine that the driving speed is abnormal. In some embodiments, when the average speed of the vehicle in a road section is not within a speed limit of the road section (e.g., lower than a lowest speed limit or higher than the highest speed limit value) , the processing device 110 may determine that the driving speed is abnormal.

In some embodiments, determining whether the current trip associated with the current order is abnormal may include identifying an abnormality type of the current trip based on the current order data and/or the real-time state data, and determining a danger level of the current trip based on the abnormality type of the current trip. For example, the processing device 110 may determine whether the vehicle deviates from the preset route based on the positioning data associated with the vehicle and the preset route in the current order data. The processing device 110 may determine the danger level of the current trip based on a deviation degree (i.e., a distance between a current position of the vehicle and the preset route) . As another example, the processing device 110 may determine whether the vehicle deviates from the preset route based on the positioning data associated with the vehicle and the preset route. The processing device 110 may determine whether the current vehicle is in the remote area based on the remote level of the current position of the vehicle. If the vehicle deviates from the preset route and is in the remote area, the processing device 110 may determine the danger level of the current trip based on the deviation from the preset route and the remote level.

In some embodiments, the processing device 110 may determine the abnormality type and the danger level of the current trip based on historical statistical data, a function, a machine learning model, etc. For example, the processing device 110 may obtain a historical order with an abnormal trip within recent ten, five or three years. The processing device 110 may estimate the abnormality type and danger level of the current order according to a statistical rule. In some embodiments, the processing device 110 may determine the abnormality type in the current order by establishing a relationship (e.g., a function) between the current order data (and/or the real-time state data associated with the current order) and the abnormality type. The processing device 110 may determine the danger level of the current trip by establishing a relationship (e.g., a function) between the current order data (and/or the real-time state data of the current order) and the danger level of current trip. In some embodiments, the processing device 110 may determine the abnormality type in the current order based on an abnormality identification model. The abnormality identification model may be configured to determine the abnormality type of the current order. For example, the abnormality identification model may determine whether the vehicle deviates from the preset route. The abnormality identification model may determine whether the vehicle is in a remote area. The abnormality identification model may determine whether the vehicle stops abnormally in the current trip. The abnormality identification model may determine whether the driving speed of the vehicle is abnormal. The processing device 110 may determine the danger level of the current trip based on an abnormality evaluation model. In some embodiments, the danger level may be associated with an abnormality probability, a level of abnormality, or a ranking of abnormality.

In some embodiments, the processing device 110 may determine whether the current trip associated with the current order is abnormal, the abnormality type of the current trip, and/or the danger level of the current trip using a supervised learning model. For example, the processing device 110 may determine an abnormality identification model configured to determine the abnormality type of the current order. The supervised learning may refer to training or obtaining a model from existing training samples (i.e., known data and its corresponding output) to implement data discrimination or classification. In some embodiments, the supervised learning model may include a machine learning model, for example, a neural network (NN) model such as a classification, a logistic regression (Logistic Regression) model, a k-Nearest Neighbor (KNN) model, a Naive Bayes (NB) model, etc.

In some embodiments, the supervised learning model may include a sequence model, for example, a deep recurrent neural network (RNN) model. Sequence data with respect to the real-time status data in the driving process may be input into the RNN model, which can analyze and process the input data with different sequence lengths.

In some embodiments, the order related data may be input into the supervised learning model to determine whether the current trip associated with the current order is abnormal, the abnormality type of the current trip, and/or the danger level of the current trip in the driving process of the vehicle.

In some embodiments, the data input into the supervised learning model may include feature information of the order related data, for example, feature information of identity information of the service provider, vehicle identification information related to the service provider, feature information of a service time, feature information of a starting location of the current trip, feature information of a destination of the current trip, feature information of a driving route of the current trip, feature information of identity information of a service requester, feature information of positioning data of a terminal device associated with the current order, feature information of state data of the terminal device associated with the current order, feature information of positioning data associated with the vehicle, feature information of state data associated with the vehicle, feature information of environmental data inside the vehicle, feature information of environmental data around the vehicle, or the like, or any combination thereof.

In some embodiments, the supervised learning model may be determined by training a preliminary model based on a plurality of training samples associated with historical order related data in a plurality of historical driving processes of a plurality of historical vehicles. In some embodiments, one type of the historical order related data may be input to the preliminary model to determine the supervised learning model. For example, a supervised learning model may be determined by training the preliminary model based on historical feature information of historical driving speeds of the historical vehicles, and the supervised learning model trained so may be configured to determine whether a driving speed of a vehicle is abnormal. As another example, a supervised learning model may be determined by training the preliminary model based on feature information historical locations associated with the historical vehicles, and the supervised learning model trained so may be configured to determine whether the vehicle is in a remote area. As a further example, a supervised learning model may be determined by training the preliminary model based on feature information of historical stopping times and historical count of stops of the historical vehicles and the supervised learning model trained so may be configured to determine whether the vehicle stops abnormally in the current trip.

In some embodiments, two or more types of the historical status data may be input to the preliminary model to determine the supervised learning model. For example, a supervised learning model may be determined by training the preliminary model based on historical feature information of historical locations of the historical vehicles and feature information historical driving behavior associated with the historical vehicles, and the supervised learning model trained so may be configured to determine whether the vehicle is in a remote area.

In some embodiments, the training samples of the supervised learning model may be obtained based on the historical order related data, theoretical data of driving processes, theoretical data of service orders, or the like, or any combination thereof.

For example, the theoretical data of the driving processes may include a normal driving speed of a vehicle, a normal acceleration of a vehicle, a normal location of the vehicle or other data of a vehicle. The theoretical data of the service order may include a normal deviation range of a service time of the service order, a normal deviation range of a service route of the service order, or the like, or any combination thereof. In some embodiments, the theoretical data of the driving processes and the theoretical data of the service orders may determine based on empirical calculations, physical laws, public research data, etc.

In some embodiments, the processing device 110 may determine whether the current trip associated with the current order is abnormal, the abnormality type of the current trip, and/or the danger level of the current trip using an unsupervised learning model. The unsupervised learning model may refer to obtaining results by directly modeling and analyzing data without labeling the data.

In some embodiments, the unsupervised learning model may use a manner that analyzes the data after vectorization. In some embodiments, the vectorization may refer to characterizing the order related data as feature vectors. Taking the driving age of the service provider as an example, a driving year of 0-3 years may proportionally correspond to [1, 0, 0] . A driving year of 3-6 years may proportionally correspond to [0, 1, 0] . A driving year of more than 6 years may proportionally correspond to [0, 0, 1] .

In some embodiments, the unsupervised learning may perform such as cluster analysis and similarity calculation after the data vectorization. In some embodiments, feature vectors of historical status data may be clustered, and then a differences between feature information of the order related data and historical feature information may be determined, for example, determining a distance between the feature information and the center of the cluster and determining the similarity between feature information and historical feature information similar to the feature information. Based on the distance and the similarity, historical order related data that is most similar to the order related data may be analyzed. The trip condition associated with the order related data may be determined based on the historical order related data. For example, the historical order related data may indicate that a historical vehicle associated with the historical order related data is in an abnormal driving condition, and a trip associated with the order related data is in the abnormal trip condition may be determined.

In some embodiments, the cluster analysis may be configured to determine the abnormal trip condition and/or the risk information according to a degree of deviation analysis. The degree of deviation may refer to a degree of deviation of the feature information of the order related data from the feature information of the historical order related data. In some embodiments, due to the low frequency of abnormal trip events, it may be difficult to obtain sufficient samples based on the supervised learning model. The process and the accuracy for determining the degree of deviation may not depend on the count of the abnormal trip events, which makes the process for the trip condition identification in the present disclosure apply to various application scenarios.

In some embodiments, the feature information may be represented by vectors. The following may use a current vector to represent a vector of the feature information of the order related data, and use a history vector representing a vector of the historical feature information of the historical order related data. In some embodiments, the current vector may correspond to a current time point, a time period, for example, 1 second, 2 seconds, etc.

In some embodiments, the historical feature information of the historical order related may be determined in the same way as the feature information of the current data. In some embodiments, the historical feature information may be pre-determined and stored.

In some embodiments, the degree of deviation may be determined by various manners, and may also be determined by a transformation or combination of the various manners. In some embodiments, the deviation may be determined by determining an average distance between the current vector and one or more historical vectors nearest to the current vector.

In some embodiments, the deviation may be determined by determining the distance between the current vector and a clustering center of the one or more historical vectors nearest to the current vector. The clustering center of the one or more historical vectors may be pre-determined obtained by other manners, for example, a K-Means clustering technique, a mean-shift clustering technique, a density-based clustering (DBSCAN) technique, a Gaussian mixture model (GMM) , etc.

In some embodiments, one or more distances among the distances may be used as the degree of deviation. In some embodiments, the degree of deviation may be determined based on a plurality of types of current vectors. For example, the distance may be determined based on a feature vector of the current driving state and a feature vector of the current traveling trajectory feature vector to determine the degree of deviation. In some embodiments, the distance may include a Euclidean distance, a Manhattan distance, a Chebyshev distance, a Minkowski distance, a standardized Euclidean distance, a Markov distance, a Hamming distance, or the like, or any combination thereof.

In some embodiments, the processing device 110 may determine whether the trip is in the abnormal trip condition based on a relationship between at least one of the deviation degree and a degree threshold. For example, if the threshold is 0.6 and the degree of the deviation (that is, the distance) determined according to the feature of the trip is 0.9, the processing device 110 may determine that the trip is in the abnormal trip condition.

In 530, the processing device 110 (e.g., a risk response module 350) may perform a preset operation based on the abnormality judgment result.

In some embodiments, the preset operation may include an abnormality processing operation. The abnormality processing operation may include at least one of an abnormality prompt to at least one of the service provider and/or the service requester, informing a police officer, informing an emergency contact, turning on monitoring device in the vehicle, triggering a reporting mechanism in the terminal device, and contacting one or more service providers around the service requester for help. In some embodiments, the abnormality processing operation may also include triggering an in-vehicle alarm, limiting the speed of the vehicle, controlling the vehicle to such as flashing or whistling for reminding, reducing an output power of a power system of the vehicle, or the like, or any combination thereof.

In some embodiments, the processing device 110 may perform a risk verification operation. In some embodiments, the risk verification operation may be performed when the trip is in the abnormal condition. The risk verification may refer to verifying the abnormal trip condition, the risk information associated with the trip to obtain a verification result. If the verification result indicates that the abnormal trip condition does not exist, the abnormality judgment result may be adjusted. If the verification result indicates that the abnormal trip condition exists, the abnormality judgment result may be verified to be true. At least one abnormality processing operation of the abnormal trip condition may be continued to perform.

In some embodiments, vehicle remote diagnosis may also be used, such as calling a diagnostic program, for further verification. For example, a diagnostic program of a vehicle may be called to diagnose the abnormal trip conditions determined for the first time and determine whether the abnormal driving condition is accurate.

In some embodiments, if the abnormal trip condition is determined based on a part of the order related data for the first time. During the risk verification, more pieces of the order related data may be used for determining the condition of the trip associated with the current order.

More detailed descriptions of the at least one risk response operation may be found elsewhere in the present disclosure. See, for example, FIGs. 3 and 4 and descriptions thereof.

It should be noted that that the description of the process 500 is merely provided for the purposes of illustration, and not intended to limit the scope of the present disclosure. For persons having ordinary skills in the art, multiple variations and modifications may be made under the teachings of the present disclosure. However, those variations and modifications do not depart from the scope of the present disclosure.

FIG. 6 is an exemplary flowchart illustrating a process for training a machine learning model according to some embodiments of the present disclosure. In some embodiments, one or more operations in process 600 may be implemented in the system 100 shown in FIG. 1. For example, one or more operations in process 600 may be stored in the form of instructions in the storage device 130 and/or storage 270 and executed by the processing device 110.

In 610, the process device 110 (e.g., the first training module 330) may obtain a plurality of first historical orders. In some embodiments, the processing device 110 may obtain the plurality of first historical orders within a historical time period as training samples. For example, the plurality of first historical orders within the historical time period may include historical orders within one-week, historical orders within one month, etc. In some embodiments, the plurality of first historical orders may include historical orders that have been submitted by historical service requesters and recorded in the system 100, such as completed orders, halfway cancelled orders that have been started. In some embodiments, the plurality of first historical orders may be obtained from one or more components of the system 100 (e.g., the storage device 130, the server 110, the service requester terminal 120, and the information source 160) via the network 150.

In 620, the processing device 110 (e.g., the first training module 330) may obtain historical order data associated with the plurality of first historical orders and/or historical real-time state data associated with the plurality of first historical orders. In some embodiments, the historical order data associated with the plurality of first historical orders may include historical identity information of a service provider associated with each first historical order, historical vehicle identification information related to the service provider associated with each first historical order, a historical service time associated with associated with each first historical order, a historical starting location of a trip associated with each first historical order, a historical destination of the trip associated with each first historical order, a historical driving route of the trip associated with each first historical order, historical identity information of a service requester associated with each first historical order, or the like, or any combination thereof. In some embodiments, the historical real-time state data associated with the plurality of first historical orders may include historical positioning data of a terminal device associated with each first historical order, historical state data of a vehicle associated with each first historical order, historical environmental data inside the vehicle associated with each first historical order, historical environmental data around the vehicle associated with each first historical order, or the like, or any combination thereof.

In 630, the processing device 110 (e.g., the first training module 330) may label at least one abnormal trip associated with the plurality of first historical orders as a positive sample, and at least one normal trip associated with the plurality of first historical orders as a negative sample. For example, when the plurality of first historical orders includes an abnormal trip, the processing device 110 may determine the abnormal trip as the positive sample. When the plurality of first historical orders includes a normal trip, the processing device 110 may determine the normal trip as the negative sample. In some embodiments, the plurality of first historical orders may be labeled manually. For example, if there is an abnormal trip on February 8, 2017, and a plurality of trips associated with all historical orders on February 8, 2017 has been obtained to be used as training samples. The abnormal trip on February 8, 2017 may be labeled as a positive sample, and normal trips on the same day may be labeled as negative samples. In some embodiments, a plurality of trips associated with the plurality of first historical orders may be labeled according to historical records of the plurality of first historical orders stored in the system 100. For example, one or more trips associated with the plurality of first historical orders with malignant events (e.g., drunk driving, overspeed driving) may be labeled as positive samples, and one or more trips associated with the plurality of first historical orders without malignant events may be labeled as negative samples. In some embodiments, the positive sample may be represented with number "1" and the negative sample may be represented with number "0" .

In 640, the processing device 110 (e.g., the first training module 330) may label an abnormality type of the positive sample. In some embodiments, the processing device 110 may label the abnormality type of the positive sample. For example, the abnormality type may include deviation from a preset route, being in a remote area, abnormal stopping in the current trip, abnormal driving speed, or the like, or any combination thereof. In some embodiments, the processing device 110 may label a main abnormality type of the positive sample. For example, if the abnormality type of the positive sample includes deviation from the preset route and abnormal stopping in the current trip, and the danger level of abnormal stopping in the current trip is greater than the danger level of deviation from the preset route, the processing device 110 may label the abnormality type of the positive sample as abnormal stopping in the current trip. In some embodiments, the processing device 110 may label two or more abnormality types of the positive sample. For example, if the abnormal types of a positive sample include deviation from the preset route and being in a remote area, the processing device 110 may label two abnormality types including deviation abnormality from a preset route and being in a remote area of the positive sample.

In 650, the processing device 110 (e.g., the first training module 330) may train an abnormality identification model based on the historical order data associated with the plurality of first historical orders and/or the historical real-time state data associated with the plurality of first historical orders, and the abnormality type of the positive sample. In some embodiments, the abnormality identification module may be a classification model. In some embodiments, the abnormality identification model may be a decision tree model, including but not limited to a classification and regression Tree (CART) , an iterative dichotomiser 3 (ID3) , a C4.5 algorithm, a random forest, a chi-squared automatic interaction detection (CHAID) , a multivariate adaptive regression splines (MARS) ) , a gradient boosting machine (GBM) , or the like, or any combination thereof. In some embodiments, the processing device 110 may select one or more nodes in a decision tree based on information gain. The processing device 110 may select the one or more nodes according to a selected condition maximizing the information gain at each time. In some embodiments, the nodes in the decision tree may correspond to characteristic parameters. In some embodiments, a characteristic parameter with the maximum information gain may be selected in each node of the abnormality identification models, and a judgment condition in each node may be a classification threshold corresponding to the characteristic parameter in each node. In some embodiments, a final identification result may be obtained by inputting characteristic parameters of an order to be identified to the trained abnormality identification model according to the judgment condition of the characteristic parameter in each node.

In 660, the processing device 110 (e.g., the first training module 330) may train an abnormality evaluation model based on the historical order data associated with the plurality of first historical orders and/or the historical real-time state data associated with the plurality of first historical orders, and labeling results of the positive sample and the negative sample. In some embodiments, the labeled results may be a classification result of the positive sample and the negative sample. For example, "1" may represent the positive sample and "0" may represent the negative sample. In some embodiments, the labeled results may include the abnormality type. For example, the abnormality type of the current trip may be labeled based on the positive sample. "A" may represent deviation from a preset route, "B" may represent being in a remote area, "C" may represent abnormal stopping in the current trip, and "D" may represent abnormal driving speed. In some embodiments, the abnormality evaluation model may be a logistic regression model, for example, a linear regression model, a logistic regression model, a polynomial regression model, a stepwise regression model, a ridge regression model, a lasso regression model, an elastic net regression model, etc. In some embodiments, a validation set may be configured to validate the abnormality evaluation model during the training process, and model parameters may be adjusted according to a validation result (e.g., the model is in a under-fitting and/or over-fitting condition) to make the abnormality evaluation model reach an optimal state. The data in the validation set may be independent with the training data of the abnormality evaluation model, and there is no intersection therebetween. In some embodiments, when a preset condition is satisfied, the training process may be terminated, and the trained abnormality evaluation model may be used as the abnormality evaluation model.

In some embodiments, the processing device 110 may train the abnormality evaluation model and the abnormality identification model independently. In some embodiments, an output of the abnormality evaluation model may be a danger probability. An output of the abnormality identification model may be an abnormality type of a trip. In some embodiments, the processing device 110 may train the abnormality evaluation model and the abnormality identification model together. For example, the processing device 110 may train the two models simultaneously, without training the abnormality evaluation model and then training the abnormality identification model. In some embodiments, the plurality of first historical orders may be the same as or different from the plurality of second historical orders. For example, historical orders data similar with those of the abnormality identification model may be used as the training samples of the abnormality evaluation model. As another example, historical orders data different from those of the abnormality identification model may be used as the training samples of the abnormality evaluation model. In some embodiments, part of the plurality of first historical order may be the same as the plurality of second historical orders. In some embodiments, characteristic parameters of the abnormality evaluation model and the characteristic parameters of the abnormality identification model may be the same. In some embodiments, the characteristic parameters of the abnormality evaluation model and the characteristic parameters of the abnormality identification model may be different. In some embodiments, part of the characteristic parameters of the abnormality evaluation model may be the same as the characteristic parameters of the abnormality identification model.

In some embodiments, the trained machine learning model (e.g., the abnormality identification model and the abnormality evaluation model) may be updated from time to time, e.g., periodically or not, based on a plurality of training samples that is at least partially different from the plurality of original training samples from which the original trained machine learning model is determined. For instance, the trained machine learning model (e.g., the abnormality identification model and the abnormality evaluation model) may be updated based on training samples including new training samples that are not in the original training samples, training samples processed using the machine learning model in connection with the original trained machine learning model of a prior version, or the like, or a combination thereof. In some embodiments, the determination and/or updating of the trained machine learning model (e.g., the abnormality identification model and the abnormality evaluation model) may be performed on a processing device, while the application of the trained machine learning model may be performed on a different processing device. In some embodiments, the determination and/or updating of the trained machine learning model (e.g., the abnormality identification model and the abnormality evaluation model) may be performed on a processing device of a system different than the system 100 or a server different than a server including the processing device 110 on which the application of the trained machine learning model is performed. For instance, the determination and/or updating of the trained machine learning model (e.g., the abnormality identification model and the abnormality evaluation model) may be performed on a first system of a vendor who provides and/or maintains such a machine learning model and/or has access to training samples used to determine and/or update the trained machine learning model, while abnormality identification and abnormality evaluation based on the provided machine learning model may be performed on a second system of a client of the vendor. In some embodiments, the determination and/or updating of the trained machine learning model (e.g., the abnormality identification model and the abnormality evaluation model) may be performed online in response to a request for abnormality identification and abnormality evaluation. In some embodiments, the determination and/or updating of the trained machine learning model (e.g., the abnormality identification model and the abnormality evaluation model) may be performed offline.

It should be noted that the above description is merely provided for the purposes of illustration, and not intended to limit the scope of the present disclosure. For persons having ordinary skills in the art, multiple variations and modifications may be made under the teachings of the present disclosure. However, those variations and modifications do not depart from the scope of the present disclosure.

FIG. 7 is a flowchart illustrating an exemplary process for determining whether a vehicle deviates from a preset route according to some embodiments of the present disclosure. In some embodiments, one or more operations in process 700 may be implemented in the system 100 shown in FIG. 1. For example, one or more operations in process 700 may be stored in the form of instructions in the storage device 130 and/or storage 270 and executed by the processing device 110.

In 710, the process device 110 (e.g., the data obtaining module 310) may obtain a current position of a vehicle and a preset route to determine a distance between the current position and the preset route.

In some embodiments, the current position of the vehicle may be obtained based on a positioning technology. The positioning technology may include but not limited to a GPS satellite positioning technology, a Bluetooth positioning technology, a WI-FI network positioning technology, a BeiDou navigation satellite system (BDS) positioning technology, a mobile communication positioning technology, etc. In some embodiments, the processing device 110 may determine the current position of the vehicle based on a positioning device installed on the vehicle. In some embodiments, the processing device 110 may determine the current position of the vehicle based on a current position of a service provider (e.g., a driver) and a current position of a service requester (e.g., a passenger) . For example, when the current position of the service provider (i.e., a current position of a service provider terminal (e.g., a driver terminal) ) is the same as that of the service requester (i.e., a current position of a service requester passenger terminal (e.g., a passenger terminal) ) , the processing device 110 may determine that the current position of the service provider and/or the current position of the service requester is the current position of the vehicle.

In some embodiments, the processing device 110 may determine the preset route based on starting location information and destination information. In some embodiments, the preset route may be a planned navigation route from a starting location to a destination according to navigation. In some embodiments, the preset route may be a route with a short driving distance or a route with a less estimated driving time. In some embodiments, preset route information may include one or more navigation routes, a navigation distance, and a navigation time. In some embodiments, the distance between the current position and the preset route may be a shortest distance from the current position to the preset route.

In 720, the process device 110 (e.g., the risk determination module 320) may determine whether the vehicle deviates from the preset route based on the distance. In some embodiments, if the distance exceeds a distance threshold, the processing device 110 may determine that the vehicle deviates from the preset route. For example, if the shortest distance between the current position of the vehicle and the preset route is 150 meters, and the distance threshold is 100 meters, the processing device 110 may determine that the vehicle deviates from the preset route. In some embodiments, if the distance exceeds the distance threshold, and a duration when the vehicle deviates from the preset route exceeds a duration threshold, the processing device 110 may determine that the vehicle deviates from the preset route. For example, if a duration from a time (also referred to as a start time) when a vehicle starts to deviate from the preset route exceeds the duration threshold (e.g., 10 minutes) , the processing device 110 may determine that the vehicle deviates from the preset route. In some embodiments, the duration may be terminated at a time (also referred to as an end time) when the vehicle returns to the preset route. When the duration of the vehicle from the start time to the end time (that is, the time when the vehicle returns to the preset route) is less than a duration threshold, the processing device 110 may determine that the vehicle does not deviate from the preset route. For example, if the duration is less than 1 minute and the duration threshold is 2 minutes, the processing device 110 may determine that the vehicle does not deviate from the preset route. In some embodiments, if an accumulated duration (i.e., accumulated deviation time) exceeds the duration threshold, the processing device 110 may determine that the vehicle deviates from the preset route. In some embodiments, the processing device 110 may determine whether the vehicle deviates from the preset route based on the current position and a driving direction of the vehicle. Specifically, the navigation route may be re-planned for a user (e.g., a passenger, a driver) based on the current position of the vehicle. When the driving direction of the vehicle is not consistent with a current navigation direction, the processing device 110 may determine that vehicle deviates from the preset route.

In some embodiments, the processing device 110 may determine a deviation degree from a preset route based on the current position of the vehicle, the driving direction of the vehicle, and a preset route of the vehicle. In some embodiments, the processing device 110 may determine the deviation degree from the preset route using a supervised learning model or an unsupervised learning model. The processing device 110 may determine whether the vehicle deviates from the preset route based on the deviation degree. For example, if the deviation degree exceeds the degree threshold, the processing device 110 may determine that the vehicle deviates from the preset route.

In 730, the process device 110 (e.g., the risk determination module 320) may determine a danger level of the current trip based on the distance, at least one of current order data associated with a current order or real-time state data associated with the current order in response to a determination that the vehicle deviates from the preset route.

In some embodiments, the danger level corresponding to the deviation from the preset route may be directly determined based on the distance between the current position of the vehicle and the preset route. For example, the larger the distance is, the higher the danger level may be. In some embodiments, the processing device 110 may determine the danger level corresponding to the deviation from the preset route according to the duration when the vehicle deviates from the preset route. For example, the longer the duration is, the higher the danger level may be. In some embodiments, the processing device 110 may determine the danger level corresponding to the deviation from the preset route based on a remote degree of the current position. In some embodiments, the processing device 110 may determine the danger level corresponding to the deviation from the preset route based on a traffic condition of the driving route. For example, if the driving route is congestion, the danger level corresponding to deviation from the preset route may be relatively low. If the driving route is smooth, the danger level corresponding to deviation from the preset route may be relatively high. In some embodiments, the processing device 110 may determine the danger level corresponding to the deviation from the preset route based on a duration of executing the current order. For example, if a duration of executing the current order is from 0: 00 to 3: 00 in the morning, the danger level corresponding to deviation from the preset route may be relatively high.

FIG. 8 is a flowchart illustrating an exemplary process for determining whether a vehicle is in a remote area according to some embodiments of the present disclosure. In some embodiments, one or more operations in process 800 may be implemented in the system 100 shown in FIG. 1. For example, one or more operations in process 800 may be stored in the form of instructions in the storage device 130 and/or storage 270 and executed by the processing device 110.

In 810, the process device 110 (e.g., the risk determination module 320) may determine a remote level of a current position of a vehicle based on positioning data associated with the vehicle and a driving route of a current trip.

In some embodiments, the remote level of the current position of the vehicle may be in negative correlation with a count of orders within a distance from the current position within a time period. For example, if a location related to a current order (e.g., the starting location of the current order) appears frequently in other orders within a week, the processing device 110 may determine that the remote level of the location is relatively low, or remote levels of locations related to a driving route across the location are relatively low. As another example, if a location related to a current order (e.g., the destination of the current order) rarely appears in other orders within a month, or has not appeared in other orders before, the processing device 110 may determine that the remote level of the location is relatively high, or remote levels of locations related to a driving route across the location are relatively high. As a further example, if a location on a driving route of a current order rarely or never appears in other orders within a week, the processing device 110 may determine that remote levels of locations related to the driving route of the current order is relatively high.

In some embodiments, the remote levels of locations related to the driving route may be in negative correlation with a pedestrian volume and/or a traffic flow through a unit area region per unit time. In some embodiments, the pedestrian volume and/or the traffic flow may be divided into different levels, and different levels may correspond to different remote levels. For example, the pedestrian volume and/or the traffic flow may be divided into five levels, that is, the pedestrian volume and/or the traffic flow is 0 ～ 5 per hour, 5 ～ 100 per hour, 100 ～ 2000 per hour, 2000 ～ 5000 per hour, greater than 5000 per hour. The corresponding remote levels may include extremely remote, remote, normal, busy, and extremely busy.

In some embodiments, the remote levels of the locations related to the driving route may be determined based on traffic conditions of the locations related to the driving route. For example, the remote levels may be determined based on information obtained by traffic cameras. In some embodiments, the remote levels of the locations related to the driving route may also be determined based on video data obtained by a vehicle driving recorder and/or an in-vehicle monitoring device. In some embodiments, the remote levels of the locations related to the driving route may be updated periodically. In some embodiments, the remote levels of the locations related to the driving route may be updated in real time.

In some embodiments, the remote levels of the locations related to the driving route may be represented as a continuous value, such as a calculated value For example, the calculated value may be determined by establishing a relationship equation of a count of orders at each of the locations related to the driving route. In some embodiments, the remote levels of the locations related to the driving route may be represented as a discrete value. For example, the remote levels of the locations related to the driving route may be levels corresponding to 0, 1, 2-10. The larger the number is, the higher the remote level may be. A same range of a count of orders may correspond to one level. The remote level of each of the locations related to the driving route may be obtained according to a level corresponding to a count of orders at each of the locations related to the driving route.

In 820, the process device 110 (e.g., the risk determination module 320) may determine whether the vehicle is in a remote area based on the remote level.

In some embodiments, if the remote level exceeds a remote level threshold, the process device 110 may determine that the vehicle is in the remote area. In some embodiments, the processing device 110 may collect equally spaced sampling points on the driving route as sampling points along the driving route. For example, five equally spaced sampling points on the driving route may be collected as the sampling points along the driving route. Then the count of orders appeared in a rectangular region corresponding to each of the five sampling points along the driving route may be counted. The remote levels of the locations related to the driving route may be determined according to the count of orders appeared in each of the five sampling points along the driving route. In some embodiments, equally spaced sampling points on the driving route may be collected as the sampling points along the current route, and diffusion sampling points may be obtain based on the sampling points along the current route. For example, a sampling point on each of both sides of each sampling point along the current route may be collected as a diffusion sampling point. If there are five sampling points along the current route, ten diffusion sampling points corresponding to the five sampling points along the current route may be obtained. The count of orders corresponding to the sampling points along the current route and the diffusion sampling points may be counted to determine the remote level of the current route. If the count of orders corresponding to a sampling point along the current route or a diffusion sampling point is less than a count threshold, the remote level of the current route may be relatively high. If the count of orders corresponding to a half of the sample points along the current route or the diffusion sample points is less than a count threshold, the remote level of the current route may be extremely high.

In 830, the process device 110 (e.g., the risk determination module 320) may determine a danger level of the current trip based on the remote level, at least one of current order data associated with a current order, or real-time state data associated with the current order in response to a determination that the vehicle is in the remote area.

In some embodiments, the processing device 110 may determine the danger level corresponding to being in a remote area based on the remote level of the current position of the vehicle. In some embodiments, the processing device 110 may determine the danger level corresponding to being in a remote area based on a duration of executing the current order. For example, if a current order is executed in the middle of the night, the danger level may be relatively high. In some embodiments, the processing device 110 may determine the danger level corresponding to being in a remote area based on a duration of the vehicle being in the remote area. In some embodiments, the processing device 110 may obtain behavior characteristics of a service provider or a service requester (e.g., a passenger, a driver) to determine the danger level corresponding to being in remote areas. For example, when an alarm raised by a passenger or a driver is detected, the danger level may be relatively high. In some embodiments, the processing device 110 may determine the danger level corresponding to being in the remote area based on a credit value of the service provider or the service requester (e.g., a passenger, a driver) . For example, if the credit value of the service provider (e.g., a driver) is relatively low, the danger level may be relatively high.

FIG. 9 is a flowchart illustrating an exemplary process for determining whether a vehicle stops abnormally in a current trip according to some embodiments of the present disclosure. In some embodiments, one or more operations in process 900 may be implemented in the system 100 shown in FIG. 1. For example, one or more operations in process 900 may be stored in the form of instructions in the storage device 130 and/or storage 270, and called and/or executed by the processing device 110.

In 910, the process device 110 (e.g., the risk determination module 320) may determine whether a vehicle stops abnormally in a current trip based on a count of stops in the current trip and/or a stopping time in the current trip. In some embodiments, the processing device 110 may determine whether the vehicle stops in the current trip based on sensor data obtained from a service requester terminal (e.g., a passenger terminal) , a service provider terminal (e.g., a driver terminal) , and/or a vehicle. For example, the processing device 110 may determine a driving speed of the vehicle based on speed sensor data. If the driving speed of the vehicle is less than a speed threshold, the processing device 110 may determine that the vehicle stops in the current trip. In some embodiments, the processing device 110 may determine whether the vehicle stops based on positioning information of the service requester terminal (e.g., a passenger terminal) , the service provider terminal (e.g., a driver terminal) , and/or the vehicle. For example, a positioning device may upload the positioning information every 1 second. When the positioning information uploaded by the vehicle remains unchanged for a time period, the processing device 110 may determine that the vehicle stops in the current trip. In some embodiments, the processing device 110 may determine whether the vehicle stops based on data obtained from a driving recorder and/or a vehicle monitoring device. For example, the processing device 110 may analyze whether the environment outside the vehicle changes based on image frames. In some embodiments, the stopping time may be a duration when the driving speed is less than the speed threshold. Specifically, in a deceleration phase, a time when the driving speed of the vehicle is equal to the speed threshold may be used as a start time of the stopping time. In an acceleration phase, a time when the driving speed of the vehicle is equal to the speed threshold may be used as an end time of the stopping time. The stopping time may be a duration from the start time to the end time. In some embodiments, the stopping time may be a duration of uploading the same positioning information. In some embodiments, the stopping time may be a duration when an image associated with the environment outside the vehicle remains unchanged.

In some embodiments, if the stopping time of the vehicle exceeds the speed threshold, the processing device110 may determine that the vehicle stops abnormally. In some embodiments, if a stopping time in a single stop exceeds a single stopping threshold, the processing device may determine that the vehicle stops abnormally in the current trip. In some embodiments, when a total stopping time exceeds a total stopping threshold, the processing device 110 may determine that the vehicle stops abnormally in the current trip. In some embodiments, when a distance between two positions of the vehicle where the vehicle stops successively is less than a distance threshold, the processing device 110 may determine that the vehicle stops abnormally in the current trip. In some embodiments, the processing device 110 may determine whether the vehicle stops abnormally in the current trip based on a remote level of a stopping position of the vehicle.

In 920, the process device 110 (e.g., the risk determination module 320) may determine a danger level of a current trip based on the count of stops in the current trip and/or the stopping time in the current trip, at least one of current order data associated with a current order, or real-time state data associated with the current order in response to a determination that the vehicle stops abnormally in the current trip.

In some embodiments, the processing device 110 may determine the danger level of the current trip based on the count of stops in the current trip. In some embodiments, the processing device 110 may determine the danger level of the current trip based on the stopping time in the current trip. In some embodiments, the processing device 110 may determine the danger level of the current trip based on the remote level of the stopping position. In some embodiments, the processing device 110 may determine the danger level of the current trip based on a period of the stopping time. For example, if the period of the stopping time is between 23: 00 and 24: 00 at night, the danger level of the current trip may be relatively high.

In some embodiments, the processing device 110 may determine the danger level based on positioning information of the service provider (e.g., a driver) and the service requester (e.g., a passenger) during the stopping. For example, the processing device 110 may determine whether a driver trails a passenger. In some embodiments, the processing device 110 may determine the danger level of the current trip based on a behavior characteristic of the service provider (e.g., a driver) or the service requester (e.g., a passenger) . For example, the processing device 110 may determine whether a passenger calls other people, whether an alarm is raised, etc. In some embodiments, the processing device 110 may determine the danger level of the current trip based on a credit value of the service provider (e.g., a driver) or the service requester (e.g., a passenger) .

FIG. 10 is a flowchart illustrating an exemplary process for determining whether a driving speed of a vehicle is abnormal according to some embodiments of the present disclosure. In some embodiments, one or more operations in process 1000 may be implemented in the system 100 shown in FIG. 1. For example, one or more operations in process 1000 may be stored in the form of instructions in the storage device 130 and/or storage 270, and called and/or executed by the processing device 110.

In 1010, the process device 110 (e.g., the risk determination module 320) may determine whether a driving speed is abnormal based on the driving speed of a vehicle. In some embodiments, driving speed information may include a driving speed of the vehicle, a driving direction of the vehicle, etc. In some embodiments, the driving speed information may include a speed curve of the vehicle. In some embodiments, the driving speed information may further include an average driving speed of the vehicle over a time period. In some embodiments, the processing device 110 may obtain the driving speed information via an in-vehicle speed sensor and a user terminal (e.g., a service provider terminal) . In some embodiments, the processing device 110 may obtain the driving speed information via the positioning information reported by a positioning device of the vehicle. For example, the vehicle may report positioning information every one second. Within 5 seconds, the processing device 110 may determine the driving speed information by calculating a distance between a position of the vehicle where the vehicle is at the first second and a position of the vehicle where the vehicle is at the fifth second. In some embodiments, the processing device 110 may obtain the driving speed information according to one or more other speed acquisition technologies.

In some embodiments, the abnormal driving speed may include a relatively high driving speed, a relatively low driving speed, or the like. In some embodiments, if the average driving speed for a time period exceeds a first speed threshold (also referred to as a highest speed threshold) , the processing device 110 may determine that the driving speed is abnormal (i.e., the relatively high driving speed) . In some embodiments, if the average driving speed for a time period is less than a second speed threshold (also referred to as a lowest speed threshold) , the processing device 110 may determine that the driving speed is abnormal (i.e., the relatively low driving speed) . In some embodiments, when the average driving speed of the vehicle in a road section is not within a speed limit of the road section (e.g., lower than a lowest speed limit, or higher than a highest speed limit) , the processing device 110 may determine that the driving speed is abnormal.

In 1020, the process device 110 (e.g., the risk determination module 320) may determine a danger level of a current trip based on the driving speed, at least one of current order data associated with a current order, or real-time state data associated with the current order in response to a determination that the driving speed is abnormal.

In some embodiments, the processing device 110 may determine the danger level of the current order based on the driving speed. For example, the greater a difference between the driving speed and a highest speed threshold is, the higher the danger level may be. In some embodiments, the danger level of the current trip may be determined based on a duration of the abnormal driving speed. For example, the longer the abnormal driving speed lasts, the higher the danger level may be. In some embodiments, the processing device 110 may determine the danger level of the current trip based on a remote level of a driving region where the driving speed is abnormal. In some embodiments, the processing device 110 may determine the danger level of the current trip based on a driving duration when the driving speed is abnormal. In some embodiments, the processing device 110 may determine the danger level of the current trip based on state information of the vehicle. For example, when the driving speed is high and the vehicle is old or the vehicle has recently been repaired, the danger level may be relatively high. In some embodiments, the processing device 110 may obtain a traffic condition of a road where the driving speed is abnormal, to determine the danger level of the current trip. For example, when the driving speed is slow but the road is not crowded, the danger level of the current trip may be relatively high.

It should be noted that the above description is merely provided for the purposes of illustration, and not intended to limit the scope of the present disclosure. For persons having ordinary skills in the art, multiple variations and modifications may be made under the teachings of the present disclosure. However, those variations and modifications do not depart from the scope of the present disclosure. For example, process 1000 may further include a remote level determination operation. As another example, process 1000 may further include an operation for sending reminder information to a user (e.g., a service provider, a service requester) based on the danger level of the current trip.

The beneficial effects of the embodiments of the present disclosure may include but not limited to: identifying abnormalities in a current trip associated with a current order based on order related data, evaluating danger levels of the abnormalities of the current trip to generate an abnormality judgment result, and implementing at least one response strategy based on the abnormality judgment result to ensure the safety of a user (e.g., a driver, a passenger) ; identifying types of abnormalities, and accurately determining danger levels of the abnormalities of the current trip based on environmental data and order related data; ranking orders based on different danger levels of the orders, and processing the orders based on the ranking to improve the processing efficiency of abnormal orders. It should be noted that different embodiments may have different beneficial effects. In different embodiments, the possible beneficial effects may be any one or a combination of the foregoing and may be any other beneficial effects that may be obtained.

FIG. 11 is a schematic diagram illustrating exemplary hardware and software components of a computing device 1100 on which the processing device 110, and/or the user terminal 120 may be implemented according to some embodiments of the present disclosure. For example, the processing device 110 may be implemented on the computing device 1100 and configured to perform functions of the processing device 110 disclosed in this disclosure.

The computing device 1100 may be configured to implement the system 100 for the present disclosure. The computing device 1100 may be configured to implement any component of the system 100 that performs one or more functions disclosed in the present disclosure. For example, the processing device 110 may be implemented on the computing device 1100, via its hardware, software program, firmware, or a combination thereof. Although only one such computer is shown, for convenience, the computer functions relating to the online-to-offline service as described herein may be implemented in a distributed fashion on a number of similar platforms, to distribute the processing load.

The computing device 1100, for example, may include COM ports 1150 connected to and from a network connected thereto to facilitate data communications. The COM port 1150 may be any network port or data exchange port to facilitate data communications. The computing device 1100 may also include a processor (e.g., the processor 1120) , in the form of one or more processors (e.g., logic circuits) , for executing program instructions. For example, the processor may include interface circuits and processing circuits therein. The interface circuits may be configured to receive electronic signals from a bus 1110, wherein the electronic signals encode structured data and/or instructions for the processing circuits to process. The processing circuits may conduct logic calculations, and then determine a conclusion, a result, and/or an instruction encoded as electronic signals. The processing circuits may also generate electronic signals including the conclusion or the result (e.g., a risk determining result) and a triggering code. In some embodiments, the trigger code may be in a format recognizable by an operation system (or an application installed therein) of an electronic device (e.g., the user terminal 120) in the system 100. For example, the trigger code may be an instruction, a code, a mark, a symbol, or the like, or any combination thereof, that can activate certain functions and/or operations of a mobile phone or let the mobile phone execute a preset program (s) . In some embodiments, the trigger code may be configured to rend the operation system (or the application) of the electronic device to generate a presentation of the conclusion or the result (e.g., a risk determining result) on an interface of the electronic device. Then the interface circuits may send out the electronic signals from the processing circuits via the bus 1110.

The exemplary computing device may include the internal communication bus 1110, program storage and data storage of different forms including, for example, a disk 1170, and a read-only memory (ROM) 1130, or a random access memory (RAM) 1140, for various data files to be processed and/or transmitted by the computing device. The exemplary computing device may also include program instructions stored in the ROM 1130, RAM 1140, and/or other type of non-transitory storage medium to be executed by the processor 1120. The methods and/or processes of the present disclosure may be implemented as the program instructions. The exemplary computing device may also include operation systems stored in the ROM 1130, RAM 1140, and/or other type of non-transitory storage medium to be executed by the processor 1120. The program instructions may be compatible with the operation systems for providing the online-to-offline service. The computing device 1100 also includes an I/O component 1160, supporting input/output between the computer and other components. The computing device 1100 may also receive programming and data via network communications.

Merely for illustration, only one processor is illustrated in FIG. 11. Multiple processors are also contemplated; thus, operations and/or method steps performed by one processor as described in the present disclosure may also be jointly or separately performed by the multiple processors. For example, if in the present disclosure the processor of the computing device 1100 executes both step A and step B, it should be understood that step A and step B may also be performed by two different processors jointly or separately in the computing device 1100 (e.g., the first processor executes step A and the second processor executes step B, or the first and second processors jointly execute steps A and B) .

Having thus described the basic concepts, it may be rather apparent to those skilled in the art after reading this detailed disclosure that the foregoing detailed disclosure is intended to be presented by way of example only and is not limiting. Various alterations, improvements, and modifications may occur and are intended to those skilled in the art, though not expressly stated herein. These alterations, improvements, and modifications are intended to be suggested by this disclosure, and are within the spirit and scope of the exemplary embodiments of this disclosure.

Moreover, certain terminology has been configured to describe embodiments of the present disclosure. For example, the terms “one embodiment, ” “an embodiment, ” and/or “some embodiments” mean that a particular feature, structure or characteristic described in connection with the embodiment is included in at least one embodiment of the present disclosure. Therefore, it is emphasized and should be appreciated that two or more references to “an embodiment, ” “one embodiment, ” or “an alternative embodiment” in various portions of this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures or characteristics may be combined as suitable in one or more embodiments of the present disclosure.

Further, it will be appreciated by one skilled in the art, aspects of the present disclosure may be illustrated and described herein in any of a number of patentable classes or context including any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof. Accordingly, aspects of the present disclosure may be implemented entirely hardware, entirely software (including firmware, resident software, micro-code, etc. ) or combining software and hardware implementation that may all generally be referred to herein as a "block, " “module, ” “engine, ” “unit, ” “component, ” or “system. ” Furthermore, aspects of the present disclosure may take the form of a computer program product embodied in one or more computer readable media having computer readable program code embodied thereon.

A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated signal may take any of a variety of forms, including electro-magnetic, optical, or the like, or any suitable combination thereof. A computer readable signal medium may be any computer readable medium that is not a computer readable storage medium and that may communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable signal medium may be transmitted using any appropriate medium, including wireless, wireline, optical fiber cable, RF, or the like, or any suitable combination of the foregoing.

Computer program code for carrying out operations for aspects of the present disclosure may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB. NET, Python or the like, conventional procedural programming languages, such as the “C” programming language, Visual Basic, Fortran 1703, Perl, COBOL 1702, PHP, ABAP, dynamic programming languages such as Python, Ruby and Groovy, or other programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN) , or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider) or in a cloud computing environment or offered as a service such as a software as a service (SaaS) .

Furthermore, the recited order of processing elements or sequences, or the use of numbers, letters, or other designations, therefore, is not intended to limit the claimed processes and methods to any order except as may be specified in the claims. Although the above disclosure discusses through various examples what is currently considered to be a variety of useful embodiments of the disclosure, it is to be understood that such detail is solely for that purpose, and that the appended claims are not limited to the disclosed embodiments, but, on the contrary, are intended to cover modifications and equivalent arrangements that are within the spirit and scope of the disclosed embodiments. For example, although the implementation of various components described above may be embodied in a hardware device, it may also be implemented as a software-only solution-e.g., an installation on an existing server or mobile device.

Similarly, it should be appreciated that in the foregoing description of embodiments of the present disclosure, various features are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure aiding in the understanding of one or more of the various embodiments. This method of disclosure, however, is not to be interpreted as reflecting an intention that the claimed subject matter requires more features than are expressly recited in each claim. Rather, claimed subject matter may lie in less than all features of a single foregoing disclosed embodiment.

Claims

A method for identifying abnormalities, implemented on a computing device having one or more processors and one or more storage devices, the method comprising:

obtaining order related data, wherein the order related data includes current order data associated with a current order and real-time state data associated with the current order;

determining whether a current trip associated with the current order is abnormal based on the current order data and the real-time state data to generate an abnormality judgment result; and

performing a preset operation based on the abnormality judgment result.
The method of claim 1, wherein:

the current order data includes at least one of identity information of a service provider, vehicle identification information related to the service provider, a service time, a starting location of the current trip, a destination of the current trip, a driving route of the current trip, or identity information of a service requester; and

the real-time state data includes at least one of positioning data of a terminal device associated with the current order, state data of the terminal device associated with the current order, positioning data associated with the vehicle, state data associated with the vehicle, environmental data inside the vehicle, or environmental data around the vehicle.
The method of any one of claims 1-2, wherein an abnormality judgment result that the current trip associated with the current order is abnormal includes at least one of: deviation from a preset route, being in a remote area, abnormal stopping in the current trip, or abnormal driving speed.
The method of any one of claims 1-3, wherein determining whether a current trip associated with the current order is abnormal comprises:

identifying an abnormality type of the current trip based on the current order data and the real-time state data; and

determining a danger level of the current trip based on the abnormality type of the current trip.
The method of claim 4, wherein identifying the abnormality type and determining the danger level comprises:

determining a distance between a current position of the vehicle and the preset route based on positioning data associated with the vehicle and the driving route of the current trip;

determining whether the vehicle deviates from the preset route based on the distance; and

in response to a determination that the vehicle deviates from the preset route, determining the danger level of the current trip based on the distance, the current order data associated with the current order, and the real-time state data associated with the current order.
The method of claim 4, wherein identifying the abnormality type and determining the danger level comprises:

determining a remote level of the current position of the vehicle based on the positioning data associated with the vehicle and the driving route of the current trip;

determining whether the vehicle is in a remote area based on the remote level; and

in response to a determination that the vehicle is in the remote area, determining a danger level of the current trip based on the remote level, the current order data associated with the current order, and the real-time state data associated with the current order.
The method of claim 4, wherein identifying the abnormality type and determining the danger level comprises:

determining whether the vehicle stops abnormally in the current trip based on at least one of a count of stops in the current trip or a stopping time in the current trip; and

in response to a determination that the vehicle stops abnormally in the current trip, determining a danger level of the current trip based on the count of stops in the current trip or the stopping time in the current trip, the current order data associated with the current order, and the real-time state data associated with the current order.
The method of claim 4, wherein identifying the abnormality type and determining the danger level comprises:

determining whether a driving speed is abnormal based on the driving speed of the vehicle; and

in response to a determination that the driving speed is abnormal, determining a danger level of the current trip based on the driving speed, the current order data associated with the current order, and the real-time state data associated with the current order.
The method of claim 4, wherein identifying an abnormality type of the current trip comprises:

obtaining an abnormality identification model; and

determining the abnormality type of the current trip based on the abnormality identification model, the current order data, and the real-time state data.
The method of claim 9, wherein obtaining an abnormality identification model comprises:

obtaining a plurality of first historical orders;

obtaining historical order data associated with the plurality of first historical orders and historical real-time state data associated with the plurality of first historical orders;

labelling at least one abnormal trip associated with the plurality of first historical orders as a positive sample, and at least one normal trip associated with the plurality of first historical orders as a negative sample;

labelling an abnormality type of the positive sample; and

training the abnormality identification model based on the historical order data associated with the plurality of first historical orders, the historical real-time state data associated with the plurality of first historical orders, and the abnormality type of the positive sample.
The method of claim 4, wherein determining a danger level of the current trip based on the abnormality type of the current trip comprises:

obtaining an abnormality evaluation model; and

determining the danger level of the current trip based on the abnormal evaluation model and the abnormality type of the trip.
The method of claim 11, wherein obtaining an abnormality evaluation model comprises:

obtaining a plurality of second historical orders;

obtaining historical order data associated with the plurality of second historical orders and historical real-time state data associated with the plurality of second historical orders;

labelling at least one abnormal trip associated with the plurality of second historical orders as a positive sample, and at least one normal trip associated with the plurality of second historical orders as a negative sample; and

training the abnormality evaluation model based on the historical order data associated with the plurality of second historical orders, the historical real-time state data associated with the plurality of second historical orders, the positive sample, and the negative sample.
The method of any one of claims 1-12, wherein the preset operation includes an abnormality processing operation, and the abnormality processing operation includes at least one of:

performing an abnormality prompt to at least one of the service provider or the service requester,

informing a police officer,

informing an emergency contact,

turning on a monitoring device in the vehicle,

triggering a reporting mechanism in the terminal device, or

contacting one or more service providers around the service requester for help.
A system for identifying abnormalities, comprising:

a data obtaining module configured to obtain order related data, wherein the order related data includes current order data associated with a current order and real-time state data associated with the current order;

a risk determination module configured to determine whether a current trip associated with the current order is abnormal based on the current order data and the real-time state data to generate an abnormality judgment result; and

a risk response module configured to perform a preset operation based on the abnormality judgment result.
The system of claim 14, wherein:

the current order data includes at least one of identity information of a service provider, vehicle identification information related to the service provider, a service time, a starting location of the current trip, a destination of the current trip, a driving route of the current trip, or identity information of a service requester; and

the real-time state data includes at least one of positioning data of a terminal device associated with the current order, state data of the terminal device associated with the current order, positioning data associated with the vehicle, state data associated with the vehicle, environmental data inside the vehicle, or environmental data around the vehicle.
The system of any one of claims 14-15, wherein an abnormality judgment result that the current trip associated with the current order is abnormal includes at least one of: deviation from a preset route, being in a remote area, abnormal stopping in the current trip, or abnormal driving speed.
The system of any one of claims 14-16, wherein the risk determination module is further configured to:

identify an abnormality type of the current trip based on the current order data and the real-time state data; and

determine a danger level of the current trip based on the abnormality type of the current trip.
The system of claim 17, wherein the risk determination module is further configured to:

determine a distance between a current position of the vehicle and the preset route based on positioning data associated with the vehicle and the driving route of the current trip;

determine whether the vehicle deviates from the preset route based on the distance; and

in response to a determination that the vehicle deviates from the preset route, determine the danger level of the current trip based on the distance, the current order data associated with the current order, and the real-time state data associated with the current order.
The system of claim 17, wherein the risk determination module is further configured to:

determine a remote level of the current position of the vehicle based on the positioning data associated with the vehicle and the driving route of the current trip;

determine whether the vehicle is in a remote area based on the remote level; and

in response to a determination that the vehicle is in the remote area, determine a danger level of the current trip based on the remote level, the current order data associated with the current order, and the real-time state data associated with the current order.
The system of claim 17, wherein the risk determination module is further configured to:

determine whether the vehicle stops abnormally in the current trip based on at least one of a count of stops in the current trip or a stopping time in the current trip;

in response to a determination that the vehicle stops abnormally in the current trip, determine a danger level of the current trip based on the count of stops in the current trip or the stopping time in the current trip, the current order data associated with the current order, and the real-time state data associated with the current order.
The system of claim 17, wherein the risk determination module is further configured to:

determine whether a driving speed is abnormal based on the driving speed of the vehicle; and

in response to a determination that the driving speed is abnormal, determine a danger level of the current trip based on the driving speed, the current order data associated with the current order, and the real-time state data associated with the current order.
The system of claim 17, wherein the risk determination module is further configured to:

obtain an abnormality identification model; and

determine the abnormality type of the current trip based on the abnormality identification model, the current order data, and the real-time state data.
The system of claim 22, further comprising a first training module configured to:

obtain a plurality of first historical orders;

obtain historical order data associated with the plurality of first historical orders and historical real-time state data associated with the plurality of first historical orders;

label at least one abnormal trip associated with the plurality of first historical orders as a positive sample, and at least one normal trip associated with the plurality of first historical orders as a negative sample;

label an abnormality type of the positive sample; and

train the abnormality identification model based on the historical order data associated with the plurality of first historical orders, the historical real-time state data associated with the plurality of first historical orders, and the abnormality type of the positive sample.
The system of claim 17, wherein the risk determination module is further configured to:

obtain an abnormality evaluation model; and

determine the danger level of the current trip based on the abnormal evaluation model and the abnormality type of the trip.
The system of claim 24, further comprising a second training module configured to:

obtain a plurality of second historical orders;

obtain historical order data associated with the plurality of second historical orders and historical real-time state data associated with the plurality of second historical orders;

label at least one abnormal trip associated with the plurality of second historical orders as a positive sample, and at least one normal trip associated with the plurality of second historical orders as a negative sample; and

train the abnormality evaluation model based on the historical order data associated with the plurality of second historical orders, the historical real-time state data associated with the plurality of second historical orders, the positive sample, and the negative sample.
The system of any one of claims 14-25, wherein the risk response module is further configured to perform an abnormality processing operation, and the abnormality processing operation includes at least one of:

performing an abnormality prompt to at least one of the service provider or the service requester,

informing a police officer,

informing an emergency contact,

turning on a monitoring device in the vehicle,

triggering a reporting mechanism in the terminal device, or

contacting one or more service providers around the service requester for help.
A system comprising:

at least one storage device storing a set of instructions; and

at least one processor in communication with the at least one storage device, when executing the stored set of instructions, the at least one processor causes the system to perform operations including:

obtaining order related data, wherein the order related data includes current order data associated with a current order and real-time state data associated with the current order;

determining whether a current trip associated with the current order is abnormal based on the current order data and the real-time state data to generate an abnormality judgment result; and

performing a preset operation based on the abnormality judgment result.
The system of claim 27, wherein:

the current order data includes at least one of identity information of a service provider, vehicle identification information related to the service provider, a service time, a starting location of the current trip, a destination of the current trip, a driving route of the current trip, or identity information of a service requester; and

the real-time state data includes at least one of positioning data of a terminal device associated with the current order, state data of the terminal device associated with the current order, positioning data associated with the vehicle, state data associated with the vehicle, environmental data inside the vehicle, or environmental data around the vehicle.
The system of any one of claims 27-28, wherein an abnormality judgment result that the current trip associated with the current order is abnormal includes at least one of: deviation from a preset route, being in a remote area, abnormal stopping in the current trip, or abnormal driving speed.
The system of any one of claims 27-29, wherein to determine whether a current trip associated with the current order is abnormal, the at least one processor causes the system to perform operations including:

identifying an abnormality type of the current trip based on the current order data and the real-time state data; and

determining a danger level of the current trip based on the abnormality type of the current trip.
The system of claim 30, wherein to identify the abnormality type and to determine the danger level, the at least one processor causes the system to perform operations including:

determining a distance between a current position of the vehicle and the preset route based on positioning data associated with the vehicle and the driving route of the current trip;

determining whether the vehicle deviates from the preset route based on the distance; and

in response to a determination that the vehicle deviates from the preset route, determining the danger level of the current trip based on the distance, the current order data associated with the current order, and the real-time state data associated with the current order.
The system of claim 30, wherein to identify the abnormality type and to determine the danger level, the at least one processor causes the system to perform operations including:

determining a remote level of the current position of the vehicle based on the positioning data associated with the vehicle and the driving route of the current trip;

determining whether the vehicle is in a remote area based on the remote level; and

in response to a determination that the vehicle is in the remote area, determining a danger level of the current trip based on the remote level, the current order data associated with the current order, and the real-time state data associated with the current order.
The system of claim 30, wherein to identify the abnormality type and to determine the danger level, the at least one processor causes the system to perform operations including:

determining whether the vehicle stops abnormally in the current trip based on at least one of a count of stops in the current trip or a stopping time in the current trip; and

in response to a determination that the vehicle stops abnormally in the current trip, determining a danger level of the current trip based on the count of stops in the current trip or the stopping time in the current trip, the current order data associated with the current order, and the real-time state data associated with the current order.
The system of claim 30, wherein to identify the abnormality type and to determine the danger level, the at least one processor causes the system to perform operations including:

determining whether a driving speed is abnormal based on the driving speed of the vehicle; and

in response to a determination that the driving speed is abnormal, determining a danger level of the current trip based on the driving speed, the current order data associated with the current order, and the real-time state data associated with the current order.
The system of claim 30, wherein to identify an abnormality type of the current trip, the at least one processor causes the system to perform operations including:

obtaining an abnormality identification model; and

determining the abnormality type of the current trip based on the abnormality identification model, the current order data, and the real-time state data.
The system of claim 35, wherein to obtain an abnormality identification model, the at least one processor causes the system to perform operations including:

obtaining a plurality of first historical orders;

obtaining historical order data associated with the plurality of first historical orders and historical real-time state data associated with the plurality of first historical orders;

labelling at least one abnormal trip associated with the plurality of first historical orders as a positive sample, and at least one normal trip associated with the plurality of first historical orders as a negative sample;

labelling an abnormality type of the positive sample; and

training the abnormality identification model based on the historical order data associated with the plurality of first historical orders, the historical real-time state data associated with the plurality of first historical orders, and the abnormality type of the positive sample.
The system of claim 30, wherein to determine a danger level of the current trip based on the abnormality type of the current trip, the at least one processor causes the system to perform operations including:

obtaining an abnormality evaluation model; and

determining the danger level of the current trip based on the abnormal evaluation model and the abnormality type of the trip.
The system of claim 37, wherein to obtain an abnormality evaluation model, the at least one processor causes the system to perform operations including:

obtaining a plurality of second historical orders;

obtaining historical order data associated with the plurality of second historical orders and historical real-time state data associated with the plurality of second historical orders;

labelling at least one abnormal trip associated with the plurality of second historical orders as a positive sample, and at least one normal trip associated with the plurality of second historical orders as a negative sample; and

training the abnormality evaluation model based on the historical order data associated with the plurality of second historical orders, the historical real-time state data associated with the plurality of second historical orders, the positive sample, and the negative sample.
The system of any one of claims 27-38, wherein the preset operation includes an abnormality processing operation, and the abnormality processing operation includes at least one of:

performing an abnormality prompt to at least one of the service provider or the service requester,

informing a police officer,

informing an emergency contact,

turning on a monitoring device in the vehicle,

triggering a reporting mechanism in the terminal device, or

contacting one or more service providers around the service requester for help.
A non-transitory computer-readable storage medium, wherein

the non-transitory computer-readable storage medium includes instructions and when a computing device reads the instructions, the computing device performs a method for identifying abnormalities of any one of claims 1-13.
A device for identifying abnormalities, comprising:

at least one storage device configured to store a set of instructions; and

at least one processor configured to execute the set of instructions and perform a method for identifying abnormalities of any one of claims 1-13.