CN114771534A

CN114771534A - Control method, training method, vehicle, device, and medium for automatically driving vehicle

Info

Publication number: CN114771534A
Application number: CN202210565672.1A
Authority: CN
Inventors: 杨煌荣; 张宽
Original assignee: Apollo Intelligent Technology Beijing Co Ltd
Current assignee: Apollo Intelligent Technology Beijing Co Ltd
Priority date: 2022-05-20
Filing date: 2022-05-20
Publication date: 2022-07-22

Abstract

The disclosure provides a control method of an automatic driving vehicle, a training method and device of a deep learning model, the automatic driving vehicle, electronic equipment, a storage medium and a program product, and relates to the technical field of artificial intelligence, in particular to the technical fields of automatic driving, intelligent transportation, high-precision maps, cloud services, car networking and the like. The specific implementation scheme is as follows: in response to receiving the lane change instruction, determining a first target lane change entry from a plurality of lane change entries based on target scene data, the target scene data including data related to the plurality of lane change entries; determining a lane change planning path based on the first target lane change convergence port; and controlling the vehicle to run according to the lane-changing planned route.

Description

Control method, training method, vehicle, device, and medium for automatically driving vehicle

Technical Field

The present disclosure relates to the field of artificial intelligence technology, and more particularly, to the field of automated driving, intelligent transportation, high-precision maps, cloud services, and internet of vehicles. Control method of an autonomous vehicle, training method of a deep learning model, apparatus, autonomous vehicle, electronic device, storage medium, and program product.

Background

Vehicles operating in autonomous driving mode may free occupants, and particularly the driver, from some driving-related responsibilities. When operating in an autonomous driving mode, the vehicle may be navigated to various locations using onboard sensors, allowing the vehicle to travel with minimal human interaction or in some cases without any passengers.

The lane change running is usually running for changing lanes to a corresponding turning lane in response to an instruction for turning running or lane change running in response to an instruction for bypassing a construction section. However, how to determine a reasonable lane-changing planning path from a complex lane-changing driving scene and control a vehicle to automatically change lanes is an important embodiment of the automatic driving capability.

Disclosure of Invention

The present disclosure provides a control method of an autonomous vehicle, a training method of a deep learning model, an apparatus, an autonomous vehicle, an electronic device, a storage medium, and a program product.

According to an aspect of the present disclosure, there is provided a control method of an autonomous vehicle, including: in response to receiving the lane change instruction, determining a first target lane change sink from a plurality of lane change sinks based on target scene data, wherein the target scene data comprises data related to the plurality of lane change sinks; determining a lane change planning path based on the first target lane change convergence port; and controlling the vehicle to run along the lane change planned path.

According to another aspect of the present disclosure, there is provided a training method of a deep learning model, including: determining a training sample, wherein the training sample comprises sample scene data and a label, the sample scene data comprises data related to a plurality of sample lane change entries, the label comprises a positive sample label and a negative sample label, the positive sample label is used for indicating a first target sample lane change entry, the first target sample lane change entry comprises a successfully-entered sample lane change entry in the plurality of sample lane change entries, the negative sample label is used for indicating a second target sample lane change entry, and the second target sample lane change entry comprises a sample lane change entry other than the first target sample lane change entry in the plurality of sample lane change entries; and training the deep learning model by using the training samples to obtain the trained deep learning model.

According to another aspect of the present disclosure, there is provided a control apparatus of an autonomous vehicle, including: a first determining module, configured to determine, in response to receiving a lane change instruction, a first target lane change entry from a plurality of lane change entries based on target scene data, wherein the target scene data includes data related to the plurality of lane change entries; the second determining module is used for determining a lane change planning path based on the first target lane change convergence port; and the driving module is used for controlling the vehicle to drive by changing the lane according to the lane changing planned route.

According to another aspect of the present disclosure, there is provided a training apparatus for a deep learning model, including: the sample determination module is used for acquiring a training sample, wherein the training sample comprises sample scene data and labels, the sample scene data comprises data related to a plurality of sample lane changing convergence ports, the labels comprise positive sample labels and negative sample labels, the positive sample labels are used for indicating a first target sample lane changing convergence port, the first target sample lane changing convergence port comprises a sample lane changing convergence port which is successfully converged in the plurality of sample lane changing convergence ports, the negative sample labels are used for indicating a second target sample lane changing convergence port, and the second target sample lane changing convergence port comprises a sample lane changing convergence port which is not the first target sample lane changing convergence port in the plurality of sample lane changing convergence ports; and the training module is used for training the deep learning model by utilizing the training samples to obtain the trained deep learning model.

According to another aspect of the present disclosure, there is provided an electronic device including: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform a method according to the present disclosure.

According to another aspect of the present disclosure, there is provided a non-transitory computer readable storage medium having stored thereon computer instructions for causing a computer to perform a method as disclosed herein.

According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program which, when executed by a processor, implements a method as disclosed herein.

According to another aspect of the present disclosure, there is provided an autonomous vehicle comprising an electronic device as described in the present disclosure.

It should be understood that the statements in this section are not intended to identify key or critical features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following description.

Drawings

The drawings are included to provide a better understanding of the present solution and are not to be construed as limiting the present disclosure. Wherein:

FIG. 1 schematically illustrates an exemplary system architecture to which control methods and apparatus for an autonomous vehicle may be applied, according to an embodiment of the disclosure;

FIG. 2 schematically illustrates an application scenario of a control method of an autonomous vehicle according to an embodiment of the disclosure;

FIG. 3 schematically illustrates a flow chart of a control method of an autonomous vehicle according to an embodiment of the disclosure;

FIG. 4 schematically illustrates a scene schematic of a control method of an autonomous vehicle according to an embodiment of the disclosure;

FIG. 5 schematically illustrates a flow chart of a control method of an autonomous vehicle according to an embodiment of the disclosure;

FIG. 6 schematically shows a flow chart of a method of training a deep learning model according to an embodiment of the present disclosure;

FIG. 7 schematically illustrates a block diagram of a control apparatus of an autonomous vehicle according to an embodiment of the disclosure;

FIG. 8 schematically illustrates a block diagram of a training apparatus for deep learning models, in accordance with an embodiment of the present disclosure; and

fig. 9 schematically shows a block diagram of an electronic device adapted to implement a control method of an autonomous vehicle according to an embodiment of the disclosure.

Detailed Description

Exemplary embodiments of the present disclosure are described below with reference to the accompanying drawings, in which various details of embodiments of the present disclosure are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.

According to an embodiment of the present disclosure, there is provided a control method of an autonomous vehicle, including: in response to receiving the lane change instruction, determining a first target lane change sink from a plurality of lane change sinks based on target scene data, wherein the target scene data comprises data related to the plurality of lane change sinks; determining a lane change planning path based on the first target lane change convergence port; and controlling the vehicle to run according to the lane change planned path.

In the technical scheme of the disclosure, the collection, storage, use, processing, transmission, provision, disclosure, application and other processing of the personal information of the related user are all in accordance with the regulations of related laws and regulations, necessary confidentiality measures are taken, and the customs of the public order is not violated.

In the technical scheme of the disclosure, before the personal information of the user is acquired or collected, the authorization or the consent of the user is acquired.

Fig. 1 schematically illustrates an exemplary system architecture to which the control method and apparatus of an autonomous vehicle may be applied, according to an embodiment of the present disclosure.

It should be noted that fig. 1 is only an example of a system architecture to which the embodiments of the present disclosure may be applied to help those skilled in the art understand the technical content of the present disclosure, and does not mean that the embodiments of the present disclosure may not be applied to other devices, systems, environments or scenarios. For example, in another embodiment, an exemplary system architecture to which the control method and apparatus of an autonomous vehicle may be applied may include a vehicle-mounted terminal of the autonomous vehicle, but the vehicle-mounted terminal may implement the control method and apparatus of the autonomous vehicle provided in the embodiments of the present disclosure without interacting with a server.

As shown in fig. 1, a system architecture 100 system according to this embodiment may include an autonomous vehicle 101, a network 102, and a server 103. Autonomous vehicle 101 may be communicatively coupled to one or more servers 103 through network 102. The network 102 may be any type of network, such as a Local Area Network (LAN) that is wired or wireless, a Wide Area Network (WAN) such as the internet, a cellular network, a satellite network, or a combination thereof. The server 103 may be any type of server or cluster of servers, such as a network or cloud server, an application server, a backend server, or a combination thereof. The server may be a data analysis server, a content server, a traffic information server, a map and point of interest (MPOI) server, or a location server, among others.

Autonomous vehicle 101 may refer to a vehicle configured to operate in an autonomous mode. But is not limited thereto. The autonomous vehicle may also operate in a manual mode, in a fully autonomous mode, or in a partially autonomous mode.

Autonomous vehicle 101 may include: the vehicle-mounted terminal comprises a vehicle-mounted terminal, a vehicle control module, a wireless communication module, a user interface module and a sensing module. Autonomous vehicle 101 may also include common components included in a common vehicle, such as: engines, wheels, steering wheels, transmissions, and the like. The common components may be controlled by the vehicle terminal and the vehicle control module using a variety of communication instructions, such as: acceleration commands, deceleration commands, steering commands, braking commands, and the like.

The various modules in autonomous vehicle 101 may be communicatively coupled to each other via an interconnect, a bus, a network, or a combination thereof. For example, may be communicatively coupled to each other via a Controller Area Network (CAN) bus. The CAN bus is a vehicle bus standard designed to allow microcontrollers and devices to communicate with each other in applications without a host.

The sensing modules may include, but are not limited to, one or more cameras, Global Positioning System (GPS) units, Inertial Measurement Units (IMU), radar units, and light detection and ranging (LIDAR) units. The GPS unit may include a transceiver operable to provide information regarding the location of the autonomous vehicle. The IMU unit may sense position and orientation changes of the autonomous vehicle based on inertial acceleration. A radar unit may represent a system that utilizes radio signals to sense obstacles within the surrounding environment of an autonomous vehicle. In addition to sensing an obstacle, the radar unit may additionally sense a speed and/or a heading of the obstacle. LIDAR units may use lasers to sense obstacles in the environment in which the autonomous vehicle is located. The LIDAR unit may include, among other components, one or more laser sources, a laser scanner, and one or more detectors. The camera may include one or more devices for capturing images of the environment surrounding the autonomous vehicle. The camera may be a still camera and/or a video camera. The camera may be mechanically movable, for example, by mounting the camera on a rotating or tilting platform.

The sensing module may also include other sensors, such as: sonar sensors, infrared sensors, steering sensors, throttle sensors, brake sensors, and audio sensors (e.g., microphones). The audio sensor may be configured to collect sound from the environment surrounding the autonomous vehicle. The steering sensor may be configured to sense a steering angle of a steering wheel, wheels of an autonomous vehicle, or a combination thereof. The throttle sensor and the brake sensor sense a throttle position and a brake position of the autonomous vehicle, respectively. In some cases, the throttle sensor and the brake sensor may be integrated into an integrated throttle/brake sensor.

The vehicle control modules may include, but are not limited to, a steering unit, a throttle unit (also referred to as an acceleration unit), and a brake unit. The steering unit is used to adjust the direction or heading of the autonomous vehicle. The throttle unit is used to control the speed of the motor or engine and thus the speed and acceleration of the autonomous vehicle. The brake unit decelerates the autonomous vehicle by providing friction to decelerate the wheels or tires of the autonomous vehicle.

The wireless communication module allows communication between the autonomous vehicle and external modules, such as devices, sensors, other vehicles, and the like. For example, the wireless communication module may communicate wirelessly directly with one or more devices or via a communication network, e.g., with a server over a network. The wireless communication module may use any cellular communication network or Wireless Local Area Network (WLAN), for example, using WiFi, to communicate with another component or module. The user interface module may be part of a peripheral device implemented within the autonomous vehicle, including, for example, a keypad, a touch screen display, a microphone, a speaker, and the like.

Some or all of the functions of the autonomous vehicle 101 may be controlled or managed by the on-board terminal, particularly when operating in the autonomous mode. The in-vehicle terminal includes the necessary hardware (e.g., processors, memory, storage devices) and software (e.g., operating systems, planning and routing programs) to receive information from the sensing module, the control module, the wireless communication module, and/or the user interface module, process the received information, and generate instructions for controlling the autonomous vehicle. Alternatively, the in-vehicle terminal may be integrated with the control module.

For example, a user who is a passenger may specify a start location and a destination for a trip, e.g., via a user interface module. The vehicle-mounted terminal obtains travel related data. For example, the in-vehicle terminal may obtain the location and the travelable path from an MPOI server, which may be part of the server. The location server provides a location service and the MPOI server provides a map service. Alternatively, such locations and maps may be cached locally in a permanent storage of the in-vehicle terminal.

The in-vehicle terminal may also obtain real-time traffic information from a traffic information system or server as the autonomous vehicle moves along the travelable path. The server may be operated by a third party entity. The functions of the server may be integrated with the in-vehicle terminal. Based on the real-time traffic information, and location information, and real-time local environment data detected or sensed by the sensing module, the in-vehicle terminal may plan an optimal path and control the autonomous vehicle, e.g., via the control module, according to the planned optimal path to safely and efficiently reach the designated destination.

It should be understood that the number of autonomous vehicles, networks, and servers in fig. 1 is merely illustrative. There may be any number of autonomous vehicles, networks, and servers, as desired for implementation.

It should be noted that the sequence numbers of the respective operations in the following methods are merely used as a representation of the operations for description, and should not be construed as representing the execution order of the respective operations. The method need not be performed in the exact order shown, unless explicitly stated.

Fig. 2 schematically shows an application scenario diagram of a control method of an autonomous vehicle according to an embodiment of the present disclosure.

As shown in fig. 2, in a case where the vehicle ADC201 (i.e., an autonomous vehicle, hereinafter referred to simply as a vehicle) needs to turn to, for example, drive left, an on-board terminal mounted on the vehicle ADC201 or a server communicatively coupled to the vehicle ADC201 generates a lane change instruction, for example, an instruction for changing a lane from a straight lane to a left-turn lane. The vehicle-mounted terminal receives the target scene data collected by the sensing module in response to receiving the lane change instruction. The control method of the autonomous vehicle provided by the embodiment of the disclosure is executed based on the target scene data. But is not limited thereto. The server may also receive target scene data collected by the sensing module in response to receiving the lane change instruction, and execute the control method of the autonomous vehicle provided by the embodiment of the disclosure based on the target scene data. The following embodiments will be exemplified by using a vehicle-mounted terminal as an execution subject, and will not be described herein.

As shown in fig. 2, the target scene data may include obstacle data about obstacles OBS201, OBS202, OBS203, OBS204 on the left-turn lane. The target scene data may further include data related to a plurality of lane change entries, such as data related to lane change entries GAP201, GAP202, and GAP203, and may further include vehicle data, environmental data around the vehicle, and road traffic regulation data. The control method of the autonomous vehicle provided by the embodiment of the disclosure can be utilized to determine the first target lane change merge point from the plurality of lane change merge points. And determining a lane change planning path based on the first target lane change convergence port. And controls the vehicle ADC201 to perform lane change driving according to the lane change planned path.

By using the control method of the automatic driving vehicle provided by the embodiment of the disclosure, the lane change selectable range is changed from the lane change convergence port adjacent to the vehicle on the target lane to the plurality of lane change convergence ports on the target lane according to the vehicle state and the surrounding obstacle state, so that the lane change selectable range is expanded, the lane change driving limitation is reduced, and the lane change driving intelligence and flexibility are improved.

Fig. 3 schematically shows a flow chart of a control method of an autonomous vehicle according to an embodiment of the disclosure.

As shown in fig. 3, the method includes operations S310 to S330.

In operation S310, in response to receiving a lane change instruction, a first target lane change entry is determined from a plurality of lane change entries based on target scene data. The target scene data includes data relating to a plurality of lane change entries.

In operation S320, a lane-change planning path is determined based on the first target lane-change merge-in.

In operation S330, the vehicle is controlled to lane-change travel according to the lane-change planned path.

According to embodiments of the present disclosure, a lane change task may be completed through multiple lane change stages in response to receiving a lane change instruction. The plurality of lane change stages may include: a lane change planning phase and a lane change execution phase.

According to an embodiment of the present disclosure, the lane change planning phase may include: and a step of generating a lane change intention or generating a lane change task in response to receiving a lane change instruction. The lane change execution phase may include: and controlling the vehicle to run along the lane change according to the lane change planned route.

For example, in the case of straight-ahead driving of the vehicle, the in-vehicle terminal may enter the lane-change planning stage from the straight-ahead stage in response to receiving the lane-change instruction, determine a first target lane-change merging port from the target lane according to the target scene data, and generate the lane-change planning path based on the first target lane-change merging port. And in the lane change execution stage, controlling the vehicle to execute lane change merging operation according to the lane change planning path of the first target lane change merging port.

According to an embodiment of the present disclosure, the lane change entry may refer to: a travel space on the target lane that allows the vehicle to perform a lane change task. The driving space is larger than the moving space of the vehicle in the course of executing the lane-changing task.

According to other embodiments of the present disclosure, in the lane change planning stage, a lane change entry adjacent to the vehicle on the target lane may be used as the first target lane change entry according to a predetermined lane change rule. The lane change junction adjacent to the vehicle on the target lane can be understood as follows: and the lane change convergence port meets the preset lane change condition. Satisfying the predetermined lane change condition may refer to: and controlling the vehicle to perform lane change driving according to the lane change planned path, so that the condition of a lane change task can be completed within a preset time. The lane change planning path not only can comprise a lane change planning track, but also can comprise a safe driving speed. The predetermined time period may be 8s, but is not limited thereto as long as a time period for which safe lane change can be secured. And when the target scene data is determined not to meet the preset lane change condition, for example, the adjacent position of the vehicle is determined not to have a lane change convergence, the vehicle stays in the lane change planning stage, waits for the lane change convergence meeting the preset lane change condition, cannot be switched to the lane change execution stage from the lane change planning stage, and further cannot execute the lane change task.

According to an embodiment of the present disclosure, in a lane change planning phase, a first target lane change merge point may be determined from a plurality of lane change merge points based on target scene data. For example, in a traffic scene in which an obstacle on the target lane forms a traffic flow, not only the lane change entrance at a position adjacent to the vehicle on the target lane but also the forward lane change entrance of the vehicle and the backward lane change entrance of the vehicle on the target lane may be taken into consideration, so that the chance of lane change increases. For example, in a case where it is determined that there is no lane change entrance adjacent to the vehicle on the target lane among the plurality of lane change entrances, the first target lane change entrance may be determined from a forward lane change entrance and a backward lane change entrance among the plurality of lane change entrances. Therefore, the selection range of the variable lane route is expanded, the time for merging the lane change is expanded, the restriction of lane change driving is reduced, and the intelligence and the flexibility of the lane change driving are improved.

According to other embodiments of the present disclosure, the lane-change planning path may be adjusted according to the first target lane-change convergence port. For example, a predetermined lane change condition prescribed for lane change, a safe travel speed, or a predetermined period of time may be adjusted. For example, the safe running speed is adjusted from 60km/h to 80km/h, or the preset time is adjusted from 8s to 10s, so that a lane change planning path meeting the preset condition can be generated according to the first target lane change entry, the vehicle is further controlled to run according to the lane change planning path, and the lane change task is completed under the condition that the adjusted preset lane change condition is met.

According to the embodiment of the present disclosure, a lane change adjusting phase (prepare change lane task) may also be added after the lane change planning phase in the case that the first target lane change convergence port is determined. For example, the lane change phases are changed from a lane change planning phase and a lane change execution phase to a lane change planning phase, a lane change adjusting phase and a lane change execution phase.

According to an embodiment of the present disclosure, the lane change adjusting phase may include: and adjusting the running speed of the vehicle. For example, in a case where it is determined that the target scene data does not satisfy the predetermined lane change condition, the travel speed is adjusted so that the adjusted target scene data satisfies the predetermined lane change condition.

According to the embodiment of the disclosure, by adjusting the driving speed of the vehicle, the relative speed and the relative position between the vehicle and the obstacle on the target lane can be adjusted, so that the adjusted target scene data, for example, the adjusted target scene data related to the first target lane change merging port, meets the preset lane change condition, and can be changed from the lane change adjusting stage to the lane change generating stage.

Fig. 4 schematically shows a scene diagram of a control method of an autonomous vehicle according to an embodiment of the disclosure.

As shown in fig. 4, in response to receiving a lane change instruction for changing lanes from a straight lane to a left lane, entering a lane change planning stage, and determining target scene data. The target scene data may be used to characterize: on a target lane, for example, a left turn lane, there are obstacles OBS401, OBS402, OBS403, a first lane change inlet is provided in front of obstacle OBS401, a second lane change inlet is provided between obstacle OBS402 and obstacle OBS403, and a third lane change inlet is provided behind obstacle OBS 403.

And the vehicle-mounted terminal determines a second lane change junction as a first target lane change junction from the first lane change junction, the second lane change junction and the third lane change junction based on the target scene data.

As shown in fig. 4, the in-vehicle terminal may generate an initial lane change planned path based on the target scene data, and determine a required lane change duration. When the vehicle runs at a safe running speed according to the initial lane change planned path and the lane change duration is longer than the preset duration, it can be determined that the target scene data does not meet the preset lane change condition, and in this case, the vehicle-mounted terminal can control the vehicle ADC401 to enter the lane change adjusting phase. The magnitude of the traveling speed of the vehicle ADC401 in the traveling direction of the straight lane is adjusted so that the lane change entrance GAP402 becomes a lane change entrance adjacent to the vehicle ADC401 on the left-turn lane. And generating a lane change planning path under the condition that the adjusted target scene data meets the preset lane change condition. And entering a lane change execution stage, controlling the vehicle ADC401 to drive along a lane change planned path according to the lane change, and finally completing a lane change task.

By using the control method of the automatic driving vehicle provided by the embodiment of the disclosure, under the condition that the target scene data related to the current driving position and the first target lane change convergence port do not meet the preset lane change condition, the adjusted target scene data can meet the preset lane change condition by setting the lane change adjusting stage, and then the lane change adjusting stage is set to actively create the lane change convergence opportunity, so that the lane change convergence capability is effectively improved, and the intelligence and flexibility in the automatic driving mode are improved while the safety of the lane change process is ensured.

According to an embodiment of the present disclosure, in a case where it is determined that the target scene data does not satisfy the predetermined lane change condition, a target acceleration for adjusting the travel speed is determined. In a case where it is determined that the target acceleration satisfies a predetermined adjustment speed condition, the travel speed is adjusted in accordance with the target acceleration. In a case where it is determined that the target acceleration does not satisfy the condition for adjusting the speed, execution of the operation for adjusting the travel speed is stopped.

According to an embodiment of the present disclosure, the predetermined adjustment speed condition may include a condition of a safe travel speed, but is not limited thereto, and may also include a motion sensing condition.

For example, when the vehicle realizes the adjustment of the travel speed by adjusting the target acceleration, the vehicle accelerates in accordance with the target acceleration, and the adjusted travel speed cannot be larger than the safe travel speed. Alternatively, the vehicle is decelerated at the target acceleration, and the vehicle cannot be decelerated too much and does not satisfy the sensory condition, for example, sudden braking.

By using the control method of the automatic driving vehicle, the lane changing capability can be improved, and meanwhile, the safety and the comfort of the lane changing process are improved.

According to the embodiment of the disclosure, in the lane change adjusting stage, under the condition of adjusting the driving speed, the current scene data can be acquired in real time through the sensing module, and the vehicle-mounted terminal receives the current scene data from the sensing module. The current scene data may include target scene data in adjusting the driving speed. The vehicle-mounted terminal determines that the current scene data meet the preset lane change canceling condition, and can cancel execution of operation for adjusting the driving speed and cancel lane change under the condition that the current scene data meet the preset lane change canceling condition.

According to an embodiment of the present disclosure, satisfying the predetermined cancellation lane change condition may include not satisfying the predetermined lane change condition. For example, the space of the first target lane-changing convergence port represented by the current scene data is smaller than the activity space of the vehicle in the lane-changing process, and no other lane-changing convergence port exists on the target lane.

According to the embodiment of the disclosure, in the lane change adjusting stage, under the condition of adjusting the driving speed, the current scene data can be collected in real time through the sensing module, and the vehicle-mounted terminal receives the current scene data from the sensing module. The vehicle-mounted terminal determines a second target lane change merging port from the plurality of lane change merging ports on the basis of the current scene data. A first lane change cost value is determined based on the first target lane change importation entry. The first lane change cost value is used to characterize: and according to the cost value of lane change driving of the first target lane change merging inlet. And determining a second lane change cost value based on the second target lane change convergence port. The second lane change cost value is used to characterize: and according to the cost value of lane change driving of the second target lane change merging inlet. And determining a new target lane-changing convergence from the first target lane-changing convergence and the second target lane-changing convergence based on the first lane-changing cost value and the second lane-changing cost value. And the vehicle-mounted terminal determines the updated lane change planning path based on the new target lane change convergence port. And controlling the vehicle to run along the lane change according to the updated lane change planned path.

According to an embodiment of the present disclosure, the first lane change cost value or the second lane change cost value may include at least one of: body feel value, time effective value, security value, etc.

According to embodiments of the present disclosure, somatosensory values may be used to characterize how comfortable a body experiences. For example, in the process of stable driving of the vehicle, if the comfort degree of passengers is high, the somatosensory value is high; on the contrary, the vehicle will have a low comfort level for the passengers due to sudden braking, and the somatosensory value is low. The time-efficient value may be used to characterize driving efficiency. For example, the shorter the travel time is, the higher the aging value is. The safety value may be used to characterize driving safety. For example, when the vehicle is traveling, the safety value is high when the risk of collision between the vehicle and the surrounding obstacle is low, and conversely, the safety value is low when the risk of collision between the vehicle and the surrounding obstacle is high.

According to the embodiment of the disclosure, a first lane change planning path is determined based on a first target lane change convergence. Based on the first lane change planned path, a first lane change cost value is determined. For example, the weighted summation is based on the first body sensation value, the first time effective value and the first safety value obtained by the first lane change planning path, so as to obtain a first lane change cost value. Similarly, a second lane-change planning path is determined based on the second target lane-change merge-entry. And determining a second lane change cost value based on the second lane change planning path.

According to the embodiment of the disclosure, the first lane change cost value and the second lane change cost value can be compared, the value with a large value is used as a target cost value, and the lane change convergence port corresponding to the target cost value is used as a new target lane change convergence port.

By using the control method of the automatic driving vehicle provided by the embodiment of the disclosure, the comprehensive performances such as comfort, safety, lane changing efficiency and the like in the lane changing process can be improved by determining the cost value.

Fig. 5 schematically shows a flow chart of a control method of an autonomous vehicle according to another embodiment of the present disclosure.

As shown in fig. 5, a lane change gap selection model N520 may be utilized to determine a first target lane change entry 530 from a plurality of lane change entries based on the target scene data 520. For example, target scene data 520 related to a plurality of lane-change merging openings is input into the lane-change gap selection model N520, and a classification result indicating a first target lane-change merging opening is output.

As shown in fig. 5, the feature extraction model N510 and the path planning model N530 may also be utilized to assist in the completion of the control method. Feature data satisfying a predetermined extraction condition may be extracted from the traffic scene data 510 as the target scene data 520 using the feature extraction model N510. For example, the traffic scene data 510 may be input into the feature extraction model N510, resulting in the target scene data 520. The lane change planning path 540 may be determined using the path planning model N530. For example, the current position 550 of the vehicle and the data of the first target lane-change junction 530 may be input into the path planning model N530 to obtain the lane-change planning path 540. And finally, controlling the vehicle to run along the lane change according to the lane change planning path 540.

According to an embodiment of the present disclosure, the lane change gap selection model may include an SVM (Support Vector Machine), a neural network, and other deep learning models.

According to an embodiment of the present disclosure, the feature extraction model may be a filtering rule, but is not limited thereto, and may further include a deep learning model such as CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), and the like for feature extraction.

According to embodiments of the present disclosure, the path planning model may include a graph search method, a fast search random tree (RRT) algorithm, and the like.

According to an embodiment of the present disclosure, traffic scene data may include: time data, vehicle data, data of obstacles related to obstacles, environmental data, road traffic regulation data, etc. The feature data satisfying the predetermined extraction condition may include at least one of: data of obstacles related to lane changes, vehicle data, environmental data, and road traffic regulation data.

According to an embodiment of the present disclosure, the data of the obstacle related to lane change may include: state data and attribute data such as the size of the obstacle, the traveling speed of the obstacle, the traveling direction of the obstacle, and the traveling acceleration of the obstacle. The vehicle data may include: status data and attribute data such as the size of the vehicle, the traveling speed of the vehicle, the traveling direction of the vehicle, and the traveling acceleration of the vehicle. The environmental data may include: objective driving environment data such as weather, visibility, road mud level, road congestion condition, road construction condition and the like. The road traffic regulation data may include: subjective running rule data such as speed limit rules, irreversible running rules, and non-solid line crossing lane change rules.

According to the embodiment of the disclosure, taking the vehicle data as an example, the vehicle data in the traffic scene data may include a license plate of the vehicle, annual inspection data of the vehicle, a size of the vehicle, state data of the vehicle, and the like. After being processed by the feature extraction model, the vehicle data in the target scene data may include state data of the vehicle, a size of the vehicle, and the like.

By using the control method of the automatic driving vehicle provided by the embodiment of the disclosure, the target scene data can be extracted from a large amount of traffic scene data through a feature extraction means, so that the target scene data is simplified data, the data processing amount is reduced, and the data processing efficiency is improved.

Fig. 6 schematically shows a flowchart of a training method of a deep learning model according to an embodiment of the present disclosure.

As shown in fig. 6, the method includes operations S610 to S620.

In operation S610, a training sample is determined. The training sample comprises sample scene data and a label, the sample scene data comprises data related to a plurality of sample lane change merging inlets, the label comprises a positive sample label and a negative sample label, the positive sample label is used for indicating a first target sample lane change merging inlet, the first target sample lane change merging inlet comprises a successfully merged sample lane change merging inlet in the plurality of sample lane change merging inlets, the negative sample label is used for indicating a second target sample lane change merging inlet, and the second target sample lane change merging inlet comprises a sample lane change merging inlet, except the first target sample lane change merging inlet, in the plurality of sample lane change merging inlets.

In operation S620, the deep learning model is trained using the training samples, resulting in a trained deep learning model.

According to the embodiment of the disclosure, the trained deep learning model can be used as a lane-change gap selection model, and the determination of the first target lane-change merging port can be used as a classification problem. The method is characterized in that a first target sample lane changing convergence port and a second target sample lane changing convergence port are added in a training sample, so that the deep learning model not only learns the characteristic data of the successfully imported sample lane changing convergence port, but also learns the characteristic data of the unselected sample lane changing convergence port in the training process, and therefore the trained deep learning model is high in accuracy and robustness in the selection of the lane changing convergence port.

According to other embodiments of the present disclosure, a target sample lane change entry may be determined from a plurality of sample lane change entries according to a corresponding lane change entry selection rule, and the effect may be verified through simulation.

According to the embodiment of the disclosure, compared with the method for determining the first target lane change inlet by using the selection rule, the method for determining the first target lane change inlet by using the lane change gap selection model can ensure that the determination process is short, the efficiency is high, and the intelligence is high.

According to an embodiment of the present disclosure, for operation S610, determining a training sample may include: initial sample scene data is acquired. Feature data satisfying a predetermined extraction condition is extracted from the initial sample scene data as sample scene data.

According to an embodiment of the present disclosure, the initial sample scene data may include: closed loop data, such as vehicle lane change success data; open loop data, such as data on the success of a human driver controlling a lane change of a vehicle; and other vehicle lane change success data collected by the vehicle.

According to an embodiment of the present disclosure, the feature data satisfying the predetermined extraction condition includes at least one of: data of obstacles related to lane changes, vehicle data, environmental data, and road traffic regulation data.

By using the training method of the deep learning model provided by the embodiment of the disclosure, sample scene data can be extracted from a large amount of initial sample scene data through a feature extraction means, so that the sample scene data is simplified data, thereby reducing the data processing amount and improving the training efficiency of the deep learning model.

According to an embodiment of the disclosure, the flow operation code of the control method of the autonomous vehicle may call a flow operation code related to a training method of the deep learning model. For example, the operation code for performing the operation of extracting feature data satisfying a predetermined extraction condition from the initial sample scene data as sample scene data coincides with the operation code for performing the operation of extracting feature data satisfying a predetermined extraction condition from the traffic scene data as target scene data. The method can ensure the consistency of the training process and the application process of the lane change gap selection model.

Fig. 7 schematically shows a block diagram of a control device of an autonomous vehicle according to an embodiment of the disclosure.

As shown in fig. 7, a control device 700 for an autonomous vehicle includes: a first determination module 710, a second determination module 720, and a travel module 730.

The first determining module 710 is configured to determine a first target lane-changing merge-in point from a plurality of lane-changing merge-ins based on target scene data in response to receiving a lane-changing instruction, where the target scene data includes data related to the plurality of lane-changing merge-ins.

And a second determining module 720, configured to determine a lane change planning path based on the first target lane change entry.

And the driving module 730 is used for controlling the vehicle to drive according to the lane change planned path.

According to an embodiment of the present disclosure, the control apparatus of an autonomous vehicle further includes, before the first determining module: and an extraction module.

And the extraction module is used for extracting the characteristic data meeting the preset extraction condition from the traffic scene data to serve as the target scene data.

According to an embodiment of the present disclosure, the control apparatus of an autonomous vehicle further includes, before the second determining module: and an adjusting module.

And the adjusting module is used for adjusting the running speed under the condition that the target scene data is determined not to meet the preset lane change condition, so that the adjusted lane change scene data meets the preset lane change condition.

According to an embodiment of the present disclosure, the control apparatus of an autonomous vehicle further includes, in a case where it is determined that the target scene data does not satisfy the predetermined lane change condition, adjusting the travel speed: a third determining module and a canceling module.

And the third determining module is used for determining the current scene data. The current scene data includes target scene data in the course of adjusting the traveling speed.

And the cancellation module is used for canceling the operation of adjusting the running speed and canceling lane changing under the condition that the current scene data is determined to meet the preset lane changing canceling condition.

According to an embodiment of the present disclosure, the control apparatus of an autonomous vehicle further includes, in a case where it is determined that the target scene data does not satisfy the predetermined lane change condition, adjusting the travel speed: the device comprises a fourth determining module, a fifth determining module, a sixth determining module and an updating module.

And the fourth determining module is used for determining a second target lane change sink from the plurality of lane change sinks on the basis of the current scene data.

And the fifth determining module is used for determining the first lane change cost value based on the first target lane change convergence port. The first lane change cost value is used to characterize: and changing the lane according to the first target lane change merging port.

And the sixth determining module is used for determining a second lane change cost value based on the second target lane change convergence port. The second lane change cost value is used to characterize: and according to the cost value of lane change driving of the second target lane change merging inlet.

And the updating module is used for determining a new target lane change convergence port from the first target lane change convergence port and the second target lane change convergence port based on the first lane change cost value and the second lane change cost value so as to determine an updated lane change planning path based on the new target lane change convergence port.

According to an embodiment of the present disclosure, the predetermined lane change condition includes: and according to the safe driving speed, completing the lane changing condition within a preset time period.

According to an embodiment of the present disclosure, the adjustment module includes an acceleration adjustment unit, a speed adjustment unit.

And an acceleration adjustment unit for determining a target acceleration for adjusting the travel speed in a case where it is determined that the target scene data does not satisfy the predetermined lane change condition.

And a speed adjusting unit for adjusting the running speed according to the target driving degree in case that it is determined that the target acceleration satisfies a predetermined adjustment speed condition.

Fig. 8 schematically shows a block diagram of a training apparatus for a deep learning model according to an embodiment of the present disclosure.

As shown in fig. 8, the training apparatus 800 for deep learning model includes: a sample determination module 810, a training module 820.

The sample determination module 810 is configured to obtain a training sample, where the training sample includes sample scene data and a label, the sample scene data includes data related to a plurality of sample lane change inlets, the label includes a positive sample label and a negative sample label, the positive sample label is used to indicate a first target sample lane change inlet, the first target sample lane change inlet includes a successfully-entered sample lane change inlet of the plurality of sample lane change inlets, the negative sample label is used to indicate a second target sample lane change inlet, and the second target sample lane change inlet includes a sample lane change inlet of the plurality of sample lane change inlets other than the first target sample lane change inlet.

And the training module 820 is used for training the deep learning model by using the training sample to obtain a trained deep learning model.

According to an embodiment of the disclosure, the sample determination module: the device comprises a data acquisition unit and a data extraction unit.

A data acquisition unit for acquiring initial sample scene data.

A data extraction unit for extracting feature data satisfying a predetermined extraction condition from the initial sample scene data as sample scene data.

According to an embodiment of the present disclosure, the feature data satisfying the predetermined extraction condition includes at least one of: obstacle data, vehicle data, environmental data, and road traffic regulation data related to lane changes.

The present disclosure also provides an electronic device, a readable storage medium, an autonomous vehicle, and a computer program product according to embodiments of the present disclosure.

According to an embodiment of the present disclosure, an electronic device includes: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform a method according to an embodiment of the disclosure.

According to an embodiment of the present disclosure, a non-transitory computer-readable storage medium having stored thereon computer instructions for causing a computer to perform a method as in an embodiment of the present disclosure.

According to an embodiment of the disclosure, a computer program product comprising a computer program which, when executed by a processor, implements a method as in an embodiment of the disclosure.

According to an embodiment of the present disclosure, an autonomous vehicle is configured with the electronic device, and the electronic device is configured to implement the control method of the autonomous vehicle described in the above embodiment when executed by a processor thereof.

FIG. 9 illustrates a schematic block diagram of an example electronic device 900 that can be used to implement embodiments of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the disclosure described and/or claimed herein.

As shown in fig. 9, the apparatus 900 includes a computing unit 901 which can perform various appropriate actions and processes in accordance with a computer program stored in a Read Only Memory (ROM)902 or a computer program loaded from a storage unit 908 into a Random Access Memory (RAM) 903. In the RAM 903, various programs and data required for the operation of the device 900 can also be stored. The calculation unit 901, ROM 902, and RAM 903 are connected to each other via a bus 904. An input/output (I/O) interface 905 is also connected to bus 904.

A number of components in the device 900 are connected to the I/O interface 905, including: an input unit 906 such as a keyboard, a mouse, and the like; an output unit 907 such as various types of displays, speakers, and the like; a storage unit 908 such as a magnetic disk, optical disk, or the like; and a communication unit 909 such as a network card, a modem, a wireless communication transceiver, and the like. The communication unit 909 allows the device 900 to exchange information/data with other devices through a computer network such as the internet and/or various telecommunication networks.

The computing unit 901 may be a variety of general and/or special purpose processing components with processing and computing capabilities. Some examples of the computing unit 901 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various dedicated Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, and so forth. The computing unit 901 executes the respective methods and processes described above, such as the control method of the autonomous vehicle. For example, in some embodiments, the control method of an autonomous vehicle may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as storage unit 908. In some embodiments, part or all of the computer program may be loaded and/or installed onto device 900 via ROM 902 and/or communications unit 909. When the computer program is loaded into RAM 903 and executed by computing unit 901, one or more steps of the above described control method of an autonomous vehicle may be performed. Alternatively, in other embodiments, the computing unit 901 may be configured to perform the control method of the autonomous vehicle in any other suitable way (e.g. by means of firmware).

Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuitry, Field Programmable Gate Arrays (FPGAs), Application Specific Integrated Circuits (ASICs), Application Specific Standard Products (ASSPs), system on a chip (SOCs), Complex Programmable Logic Devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.

Program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the program codes, when executed by the processor or controller, cause the functions/operations specified in the flowchart and/or block diagram to be performed. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user may provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), Wide Area Networks (WANs), and the Internet.

The computer system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server may be a cloud server, a server of a distributed system, or a server combining a blockchain.

It should be understood that various forms of the flows shown above may be used, with steps reordered, added, or deleted. For example, the steps described in the present disclosure may be executed in parallel, sequentially, or in different orders, and are not limited herein as long as the desired results of the technical solutions disclosed in the present disclosure can be achieved.

The above detailed description should not be construed as limiting the scope of the disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may be made in accordance with design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present disclosure should be included in the scope of protection of the present disclosure.

Claims

1. A control method of an autonomous vehicle, comprising:

in response to receiving a lane change instruction, determining a first target lane change entry from a plurality of lane change entries based on target scene data, wherein the target scene data comprises data related to the plurality of lane change entries;

determining a lane change planning path based on the first target lane change convergence port; and

and controlling the vehicle to run along the lane change according to the lane change planned path.

2. The method of claim 1, further comprising, prior to said determining a first target lane-change merge-entry from a plurality of lane-change merge-entries based on target scene data:

and extracting feature data meeting a preset extraction condition from the traffic scene data as the target scene data.

3. The method of claim 1, further comprising, prior to said determining a lane-change plan path based on said first target lane-change entry:

and in the case that the target scene data is determined not to meet the preset lane change condition, adjusting the running speed so that the adjusted target scene data meets the preset lane change condition.

4. The method of claim 3, further comprising, in the event that the adjustment to the travel speed is determined that the target scene data does not satisfy a predetermined lane-change condition:

determining current scene data, wherein the current scene data comprises target scene data in the process of adjusting the driving speed; and

and under the condition that the current scene data is determined to meet the preset lane change canceling condition, canceling the operation of adjusting the driving speed and canceling the lane change.

5. The method of claim 4, further comprising, in the event that the adjustment of the travel speed is determined that the lane-change scene data does not satisfy the predetermined lane-change condition:

determining a second target lane-changing convergence from a plurality of lane-changing convergence based on the current scene data;

determining a first lane change cost value based on the first target lane change sink entry, wherein the first lane change cost value is used for characterizing: according to the cost value of the first target lane change merging entrance lane change driving;

determining a second lane change cost value based on the second target lane change convergence port, wherein the second lane change cost value is used for characterizing: according to the cost value of the second target lane change merging entrance lane change driving; and

and determining a new target lane-changing convergence port from the first target lane-changing convergence port and the second target lane-changing convergence port on the basis of the first lane-changing cost value and the second lane-changing cost value, so as to determine an updated lane-changing planning path on the basis of the new target lane-changing convergence port.

6. The method of claim 3, wherein the predetermined lane-change condition comprises: and finishing the lane change condition within a preset time according to the safe driving speed.

7. The method of claim 2, wherein the feature data satisfying a predetermined extraction condition comprises at least one of:

data of obstacles related to lane changes, vehicle data, environmental data, and road traffic regulation data.

8. The method of claim 3, wherein said adjusting the travel speed in the event that it is determined that the target scene data does not satisfy a predetermined lane change condition comprises:

determining a target acceleration for adjusting a driving speed in a case where it is determined that the target scene data does not satisfy a predetermined lane change condition;

and under the condition that the target acceleration is determined to meet the preset adjusting speed condition, adjusting the running speed according to the target driving degree.

9. A training method of a deep learning model comprises the following steps:

determining a training sample, wherein the training sample comprises sample scene data and labels, the sample scene data comprises data related to a plurality of sample lane change entries, the labels comprise positive sample labels and negative sample labels, the positive sample labels are used for indicating first target sample lane change entries, the first target sample lane change entries comprise successfully-entered sample lane change entries of the plurality of sample lane change entries, the negative sample labels are used for indicating second target sample lane change entries, and the second target sample lane change entries comprise sample lane change entries of the plurality of sample lane change entries except the first target sample lane change entry; and

and training a deep learning model by using the training samples to obtain a trained deep learning model.

10. The method of claim 9, wherein the determining training samples comprises:

acquiring initial sample scene data; and

feature data satisfying a predetermined extraction condition is extracted from the initial sample scene data as the sample scene data.

11. The method of claim 10, wherein the feature data satisfying a predetermined extraction condition comprises at least one of:

obstacle data, vehicle data, environmental data, and road traffic regulation data related to lane changes.

12. A control apparatus of an autonomous vehicle, comprising:

a first determining module, configured to determine, in response to receiving a lane change instruction, a first target lane change sink from a plurality of lane change sinks based on target scene data, where the target scene data includes data related to the plurality of lane change sinks;

the second determining module is used for determining a lane change planning path based on the first target lane change convergence port; and

and the driving module is used for controlling the vehicle to drive according to the lane change planned path.

13. The apparatus of claim 12, further comprising, prior to the first determining means:

and the extraction module is used for extracting characteristic data meeting a preset extraction condition from the traffic scene data to serve as the target scene data.

14. The apparatus of claim 12, further comprising, prior to the second determining means:

and the adjusting module is used for adjusting the driving speed under the condition that the target scene data are determined not to meet the preset lane changing condition, so that the adjusted target scene data meet the preset lane changing condition.

15. The apparatus of claim 14, further comprising, in the event that the adjustment of the travel speed in the event that the determination is that the target scene data does not satisfy a predetermined lane change condition:

a third determining module, configured to determine current scene data, where the current scene data includes target scene data in a process of adjusting the driving speed; and

and the canceling module is used for canceling the operation of adjusting the running speed and canceling lane changing under the condition that the current scene data is determined to meet the preset lane changing canceling condition.

16. The apparatus of claim 15, further comprising, in the event that the adjustment of the travel speed in the event that the determination is that the target scene data does not satisfy a predetermined lane change condition:

a fourth determining module, configured to determine a second target lane change merge point from the plurality of lane change merge points based on the current scene data;

a fifth determining module, configured to determine a first lane change cost value based on the first target lane change sink entry, where the first lane change cost value is used to characterize: according to the cost value of the first target lane change merging entrance lane change driving;

a sixth determining module, configured to determine a second lane change cost value based on the second target lane change import, where the second lane change cost value is used to characterize: according to the cost value of the second target lane change merging entrance lane change driving; and

17. The apparatus of claim 14, wherein the predetermined lane-change condition comprises: and finishing the lane change condition within a preset time according to the safe driving speed.

18. The apparatus of claim 13, wherein the feature data satisfying a predetermined extraction condition comprises at least one of:

19. The apparatus of claim 14, wherein the adjustment module comprises:

an acceleration adjustment unit configured to determine a target acceleration for adjusting a travel speed in a case where it is determined that the target scene data does not satisfy a predetermined lane change condition;

and the speed adjusting unit is used for adjusting the running speed according to the target driving degree under the condition that the target acceleration is determined to meet the preset adjusting speed condition.

20. A training apparatus for deep learning models, comprising:

a sample determination module configured to obtain a training sample, wherein the training sample comprises sample scene data and labels, the sample scene data comprises data related to a plurality of sample lane change entries, the labels comprise positive sample labels and negative sample labels, the positive sample labels are configured to indicate a first target sample lane change entry, the first target sample lane change entry comprises a successfully merged sample lane change entry of the plurality of sample lane change entries, the negative sample labels are configured to indicate a second target sample lane change entry, and the second target sample lane change entry comprises a sample lane change entry of the plurality of sample lane change entries other than the first target sample lane change entry; and

and the training module is used for training the deep learning model by using the training samples to obtain the trained deep learning model.

21. The apparatus of claim 20, wherein the sample determination module:

a data acquisition unit for acquiring initial sample scene data; and

a data extraction unit configured to extract, as the sample scene data, feature data satisfying a predetermined extraction condition from the initial sample scene data.

22. The apparatus of claim 21, wherein the feature data satisfying a predetermined extraction condition comprises at least one of:

obstacle data, vehicle data, environmental data, and road traffic regulation data related to lane change.

23. An electronic device, comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 11.

24. A non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method of any one of claims 1 to 11.

25. A computer program product comprising a computer program which, when executed by a processor, implements the method of any one of claims 1 to 11.

26. An autonomous vehicle comprising: the electronic device of claim 23.