CN112328730A

CN112328730A - Map data updating method, related device, equipment and storage medium

Info

Publication number: CN112328730A
Application number: CN202110001028.7A
Authority: CN
Inventors: 陈威志
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2021-01-04
Filing date: 2021-01-04
Publication date: 2021-02-05
Anticipated expiration: 2041-01-04
Also published as: CN112328730B

Abstract

The application discloses a map data updating method based on an artificial intelligence technology, which comprises the steps of obtaining vehicle data; generating an anchor point according to the vehicle data, wherein the electronic map comprises target original map data, and the target original map data comprises nodes and node connecting edges; generating a fused driving track according to the anchor points and the vehicle data; acquiring target map data according to the fused driving track, wherein the target map data comprises updated nodes and updated node connecting edges; and updating the target original map data into target map data. The application also provides a related device, equipment and a storage medium. According to the method and the device, anchor points can be constructed on the electronic map based on a large amount of vehicle data, the anchor points and a large amount of vehicle data are further combined to generate a fusion driving track, namely, the driving track can be optimized by using the fixed anchor points, and therefore more accurate map data can be obtained.

Description

Map data updating method, related device, equipment and storage medium

Technical Field

The present application relates to the field of artificial intelligence technologies, and in particular, to a method, a related apparatus, a device, and a storage medium for map data update.

Background

With the increase of travel of people, the number of navigation applications is gradually increased, people can easily find a destination through the navigation applications installed on the terminal equipment, and the navigation applications provide navigation information for users by using route data provided by the route database. Because of the construction and change of road networks, map data corresponding to road networks are also constantly updated.

In order to ensure that the map application client provides accurate path service for the user, the map data of the client needs to be updated in time. At present, the node position is updated by using a Global Positioning System (GPS) track, and when enough vehicles pass through the same intersection, map data in a navigation application can be updated according to the GPS track.

However, the accuracy of GPS positioning is low, and there is usually an error of several meters or even tens of meters, and therefore, there is also a large error in the node position obtained based on the GPS track update, resulting in insufficient accuracy of map data.

Disclosure of Invention

The embodiment of the application provides a map data updating method, a related device, equipment and a storage medium, anchor points can be constructed on an electronic map based on a large amount of vehicle data, and further combined with the anchor points and the large amount of vehicle data to generate a fused driving track, namely the driving track can be optimized by using the fixed anchor points, so that more accurate map data can be obtained.

In view of the above, an aspect of the present application provides a method for updating map data, including:

acquiring vehicle data, wherein the vehicle data comprises an image to be identified, a position parameter and a direction parameter;

generating anchor points according to the vehicle data;

generating a fused driving track according to the anchor points and the vehicle data;

acquiring target map data according to the fused driving track, wherein the target map data comprise updated nodes and updated node connecting edges;

and updating the target original map data in the electronic map into target map data.

Another aspect of the present application provides a map data updating apparatus, including:

the system comprises an acquisition module, a recognition module and a display module, wherein the acquisition module is used for acquiring vehicle data, and the vehicle data comprises an image to be recognized, a position parameter and a direction parameter;

the generation module is used for generating anchor points according to the vehicle data;

the generating module is also used for generating a fused driving track according to the anchor point and the vehicle data;

the acquisition module is further used for acquiring target map data according to the fused driving track, wherein the target map data comprises updated nodes and updated node connecting edges;

and the updating module is used for updating the target original map data in the electronic map into target map data.

In one possible design, in another implementation of another aspect of an embodiment of the present application,

the obtaining module is specifically configured to receive vehicle data corresponding to the vehicle if the vehicle turns before passing through the target node or after passing through the target node, where the target node is any node included in the target original map data, and the vehicle is one or more vehicles among the N vehicles.

the generating module is specifically used for acquiring an interested area according to the image to be identified included in the vehicle data;

acquiring characteristic points corresponding to the region of interest;

generating expanded characteristic information according to the characteristic points corresponding to the region of interest and the direction parameters and the position parameters included in the vehicle data;

generating three-dimensional expansion characteristic information corresponding to the region of interest according to the characteristic points corresponding to the region of interest and the expansion characteristic information;

and determining the position corresponding to the anchor point in the electronic map according to the three-dimensional expansion feature information corresponding to the region of interest.

the generating module is specifically used for acquiring an image recognition result through an image recognition model based on the image to be recognized included in the vehicle data;

if the indication mark is determined to exist according to the image recognition result, acquiring a main region of interest, wherein the main region of interest belongs to the region of interest;

and if the static reference object is determined to exist according to the image identification result, acquiring a secondary region of interest, wherein the secondary region of interest belongs to the region of interest.

the generating module is specifically used for determining the three-dimensional expansion characteristic information corresponding to the region of interest as the three-dimensional expansion characteristic information corresponding to the anchor point to be matched;

acquiring three-dimensional expansion characteristic information corresponding to the first candidate anchor point;

matching the three-dimensional extended characteristic information corresponding to the anchor point to be matched with the three-dimensional extended characteristic information corresponding to the first candidate anchor point;

if the three-dimensional expansion feature information corresponding to the anchor point to be matched is successfully matched with the three-dimensional expansion feature information corresponding to the first candidate anchor point, fusing the anchor point to be matched and the first candidate anchor point to obtain the position corresponding to the anchor point in the electronic map;

and if the three-dimensional expansion feature information corresponding to the anchor point to be matched fails to be matched with the three-dimensional expansion feature information corresponding to the first candidate anchor point, determining the anchor point to be matched as a second candidate anchor point.

the generating module is specifically used for determining the matching degree according to the three-dimensional expansion feature information corresponding to the anchor point to be matched and the three-dimensional expansion feature information corresponding to the first candidate anchor point;

after matching the three-dimensional extended feature information corresponding to the anchor point to be matched with the three-dimensional extended feature information corresponding to the first candidate anchor point, the method further comprises the following steps:

if the matching degree is greater than or equal to the threshold value of the matching degree, determining that the three-dimensional expansion feature information corresponding to the anchor point to be matched is successfully matched with the three-dimensional expansion feature information corresponding to the first candidate anchor point;

and if the matching degree is less than or equal to the threshold value of the matching degree, filtering the anchor points to be matched.

the generating module is specifically used for generating multidimensional Gaussian distribution corresponding to the electronic map according to the vehicle data;

generating a two-dimensional driving probability map corresponding to the electronic map according to the anchor points and the vehicle data;

determining a lane type according to the multidimensional Gaussian distribution, wherein the lane type is a single lane type or a multi-lane type;

and generating a fused driving track according to the lane type and the two-dimensional driving probability map.

the generating module is specifically used for constructing a two-dimensional original map corresponding to the electronic map, wherein the two-dimensional original map comprises (N multiplied by M) grids, and both N and M are integers greater than 1;

acquiring K local driving tracks according to position parameters included in the vehicle data, wherein the vehicle data include vehicle data of K vehicles, the vehicle data of each vehicle are used for generating one local driving track, and K is an integer greater than 1;

according to the anchor points, optimizing each local driving track in the K local driving tracks to obtain K optimized local driving tracks;

and mapping the K optimized local vehicle tracks to a two-dimensional original graph to obtain a two-dimensional driving probability graph, wherein the probability value corresponding to the grid in the two-dimensional driving probability graph is in positive correlation with the passing times of the K optimized local vehicle tracks.

the generating module is specifically used for acquiring the length of the main direction and the length of the secondary direction according to the multi-dimensional Gaussian distribution;

generating a flat parameter according to the length in the main direction and the length in the secondary direction;

determining a segmentation threshold value according to the flattening parameters;

and determining the lane type according to the segmentation threshold.

the generation module is specifically used for carrying out binarization processing on the two-dimensional driving probability map by adopting a segmentation threshold value to obtain a binarization driving probability map, wherein the binarization driving probability map comprises a first region, and the probability value of each pixel point in the first region is greater than or equal to the segmentation threshold value;

acquiring a second region according to the two-dimensional driving probability map, wherein the probability value of each pixel point in the second region is greater than 0;

determining an intersection ratio according to the first area and the second area;

if the intersection ratio is smaller than or equal to the intersection ratio threshold value, determining that the lane type is a single lane type;

and if the intersection ratio is greater than the intersection ratio threshold value, determining that the lane type is a multi-lane type.

if the lane type is a single lane type, performing curve fitting processing on a first area included in the binarization driving probability map to generate a fusion driving track;

the obtaining module is specifically configured to extract Q updated nodes from the fused vehicle trajectory, where Q is an integer greater than 1, and the Q updated nodes at least include nodes before turning and nodes after turning.

if the lane type is a multi-lane type, constructing a polygonal area according to a first area included in the binary driving probability map;

generating a fused driving track according to the polygonal area;

Another aspect of the present application provides a server, including: a memory, a processor, and a bus system;

wherein, the memory is used for storing programs;

the processor is used for executing the program in the memory, and the processor is used for executing the method provided by the aspects according to the instructions in the program code;

the bus system is used for connecting the memory and the processor so as to enable the memory and the processor to communicate.

Another aspect of the present application provides a computer-readable storage medium having stored therein instructions, which when executed on a computer, cause the computer to perform the method of the above-described aspects.

In another aspect of the application, a computer program product or computer program is provided, the computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions to cause the computer device to perform the method provided by the above aspects.

According to the technical scheme, the embodiment of the application has the following advantages:

the embodiment of the application provides a map data updating method, which comprises the steps of firstly obtaining vehicle data, then generating an anchor point according to the vehicle data, then generating a fused driving track according to the anchor point and the vehicle data, and finally obtaining target map data according to the fused driving track so as to update target original map data into target map data. By the aid of the method, anchor points can be constructed on the electronic map based on a large amount of vehicle data, and further the anchor points and the large amount of vehicle data are combined to generate a fused driving track, namely the driving track can be optimized by the aid of the fixed anchor points, so that more accurate map data can be obtained.

Drawings

FIG. 1 is a block diagram of a map data update system according to an embodiment of the present application;

FIG. 2 is a schematic diagram of map data in an embodiment of the present application;

FIG. 3 is a schematic diagram of a directed graph in an embodiment of the present application;

FIG. 4 is a schematic diagram of an embodiment of a map data updating method in an embodiment of the present application;

FIG. 5 is a schematic diagram of outputting pose information based on a loose coupling manner in an embodiment of the present application;

FIG. 6 is a schematic diagram of outputting pose information based on a close-coupled mode in an embodiment of the present application;

FIG. 7 is a schematic diagram illustrating an embodiment of updating target original map data into target map data;

FIG. 8 is another diagram illustrating the updating of target original map data into target map data according to an embodiment of the present application;

FIG. 9 is a schematic illustration of an advanced turn or a retarded turn in an embodiment of the present application;

FIG. 10 is a schematic diagram of a relationship between lane types and a multi-dimensional Gaussian distribution in an embodiment of the present application;

FIG. 11 is a schematic diagram of a two-dimensional original map and a two-dimensional driving probability map according to an embodiment of the present application;

FIG. 12 is a diagram of a multi-dimensional Gaussian distribution in an embodiment of the present application;

FIG. 13 is a schematic view of the major and minor lengths in the present embodiment;

FIG. 14 is a schematic illustration of the cross-over ratio between the first region and the second region in an embodiment of the present application;

FIG. 15 is a schematic diagram of extracting a fused vehicle trajectory from a polygonal region according to an embodiment of the present application;

FIG. 16 is a flowchart illustrating an overall process of a map data updating method according to an embodiment of the present application;

fig. 17 is a schematic diagram of an embodiment of a map data updating apparatus according to an embodiment of the present application;

fig. 18 is a schematic structural diagram of a server in the embodiment of the present application.

Detailed Description

The terms "first," "second," "third," "fourth," and the like in the description and in the claims of the present application and in the drawings described above, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are, for example, capable of operation in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "corresponding" and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

Navigation positioning has important application value and prospect in the fields of unmanned driving, public transportation, daily travel and the like, and civil navigation application gradually deepens into the lives of people along with the popularization of terminal equipment and the optimization of a computing platform. Navigation applications often require electronic maps (electronic maps) and support from positioning systems.

Among them, an electronic map, i.e., a digital map, is a map that is digitally stored and referred to using computer technology. The method for storing information in electronic map generally uses vector image storage, and the map scale can be enlarged, reduced or rotated without affecting the display effect. Electronic maps typically utilize geographic information systems to store and transmit map data, as well as other information systems. The electronic map is a system for map making and application, is a map generated by the control of an electronic computer, is a screen map based on a digital cartographic technology, and is a visual real map.

Positioning systems include, but are not limited to, the Global Positioning System (GPS), the BeiDou Navigation Satellite System (BDS), the Galileo Satellite Navigation System (Galileo Satellite Navigation System), and the Glonass Satellite Navigation System (Glonass Satellite Navigation System).

Navigation applications need to provide accurate navigation information for users, which requires electronic maps with higher reliability and accuracy, and based on this, rapid updating of electronic map data is a hot issue that has attracted attention in recent years. Under the condition of rapid city development, information of various objects on the surface of a city is changed very frequently, and in order to achieve express updating of map data, the method for updating the map data is provided, and is applied to a map data updating system shown in fig. 1. The server related to the application can be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, and can also be a cloud server providing basic cloud computing services such as cloud service, a cloud database, cloud computing, a cloud function, cloud storage, Network service, cloud communication, middleware service, domain name service, safety service, Content Delivery Network (CDN), big data and an artificial intelligence platform. The terminal device may be a vehicle-mounted terminal, a smart phone, a tablet computer, a notebook computer, a palm computer, a personal computer, a smart television, a smart watch, and the like, but is not limited thereto. The terminal device and the server may be directly or indirectly connected through wired or wireless communication, and the application is not limited herein. The number of servers and terminal devices is not limited.

Specifically, taking a terminal device as an example of a vehicle-mounted terminal, the vehicle terminal is arranged on a vehicle, and the vehicle terminal is provided with a positioning system and can report position parameters and direction parameters to a server in real time. Taking a terminal device as an example of a smart phone placed in a vehicle, the smart phone is provided with a positioning system, and can also report position parameters and direction parameters to a server in real time. In addition, the vehicle is also provided with a camera (camera), the shot image stream (including a plurality of frames of images to be recognized) can be reported in real time, the position parameter, the direction parameter and the image stream belong to vehicle data, the server generates an anchor point according to the received vehicle data, then generates a fusion driving track according to the anchor point and the vehicle data, generates target map data based on the fusion driving track, and updates original target original map data into the target map data. Thereby, the server can transmit the updated target map data to the terminal device. The map update is a process of correcting the map content in real-time. The method aims to reflect the actual changes of human and natural elements in time and keep the map presence, accuracy and reliability.

In the process of generating an anchor point, the content in a plurality of frames of images to be recognized needs to be recognized, and therefore, the process involves Computer Vision (CV) technology based on Artificial Intelligence (AI). Artificial intelligence is a theory, method, technique and application system that uses a digital computer or a machine controlled by a digital computer to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use the knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technique of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a manner similar to human intelligence. Artificial intelligence is the research of the design principle and the realization method of various intelligent machines, so that the machines have the functions of perception, reasoning and decision making. The artificial intelligence technology is a comprehensive subject and relates to the field of extensive technology, namely the technology of a hardware level and the technology of a software level. The artificial intelligence infrastructure generally includes technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and the like.

Computer Vision technology (CV) is a science for researching how to make a machine "see", and further refers to that a camera and a Computer are used to replace human eyes to perform machine Vision such as identification, tracking and measurement on a target, and further image processing is performed, so that the Computer processing becomes an image more suitable for human eyes to observe or transmitted to an instrument to detect. As a scientific discipline, computer vision research-related theories and techniques attempt to build artificial intelligence systems that can capture information from images or multidimensional data. The computer vision technology generally includes image processing, image Recognition, image semantic understanding, image retrieval, Optical Character Recognition (OCR), video processing, video semantic understanding, video content/behavior Recognition, three-dimensional object reconstruction, 3D technology, virtual reality, augmented reality, synchronous positioning, map construction, and other technologies, and also includes common biometric technologies such as face Recognition and fingerprint Recognition.

In addition, a large amount of vehicle data needs to be involved in the process of generating the anchor point, however, the vehicle data has certain privacy, so the vehicle data can be stored in the blockchain, and it can be understood that the blockchain is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, a consensus mechanism, an encryption algorithm, and the like. A block chain (Blockchain), which is essentially a decentralized database, is a series of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, so as to verify the validity (anti-counterfeiting) of the information and generate a next block. The blockchain may include a blockchain underlying platform, a platform product services layer, and an application services layer.

The block chain underlying platform can comprise processing modules such as user management, basic service, intelligent contract and operation monitoring. The user management module is responsible for identity information management of all blockchain participants, and comprises public and private key generation maintenance (account management), key management, user real identity and blockchain address corresponding relation maintenance (authority management) and the like, and under the authorization condition, the user management module supervises and audits the transaction condition of certain real identities and provides rule configuration (wind control audit) of risk control; the basic service module is deployed on all block chain node equipment and used for verifying the validity of the service request, recording the service request to storage after consensus on the valid request is completed, for a new service request, the basic service firstly performs interface adaptation analysis and authentication processing (interface adaptation), then encrypts service information (consensus management) through a consensus algorithm, transmits the service information to a shared account (network communication) completely and consistently after encryption, and performs recording and storage; the intelligent contract module is responsible for registering and issuing contracts, triggering the contracts and executing the contracts, developers can define contract logics through a certain programming language, issue the contract logics to a block chain (contract registration), call keys or other event triggering and executing according to the logics of contract clauses, complete the contract logics and simultaneously provide the function of upgrading and canceling the contracts; the operation monitoring module is mainly responsible for deployment, configuration modification, contract setting, cloud adaptation in the product release process and visual output of real-time states in product operation, such as: alarm, monitoring network conditions, monitoring node equipment health status, and the like.

The platform product service layer provides basic capability and an implementation framework of typical application, and developers can complete block chain implementation of business logic based on the basic capability and the characteristics of the superposed business. The application service layer provides the application service based on the block chain scheme for the business participants to use.

In view of the present disclosure, reference is made to certain terms, which will be referred to separately below.

1. A plurality of sensors: the system mainly comprises a GPS (global positioning system), an Inertial Measurement Unit (IMU), a camera, a wheel speed sensor, a magnetometer (magnetometer), and the like. When an electronic map is constructed, a pure laser radar sensor cannot well complete map construction work, at this time, different sensors need to be considered, various data are obtained to perform fusion of the sensors, pose correction is achieved, and map construction under a complex large-scale scene is finally completed.

2. GPS: the position parameters can be obtained based on GPS positioning, and the basic principle of GPS positioning is to determine the position of a point to be measured by adopting a space distance rear intersection method according to the known calculation data of the instantaneous position of a satellite moving at a high speed. Assuming that a GPS receiver is arranged on the ground point to be measured at the time t, the time of a GPS signal reaching the receiver can be measured, and the longitude and latitude where the point to be measured is located can be determined by the satellite ephemeris and other data received by the receiver.

3. An IMU: the orientation parameters are obtained based on an IMU, which is a device that measures the three-axis attitude angle (or angular velocity) and acceleration of an object, and is capable of measuring the current acceleration and rotation speed, and the like. Gyroscopes and accelerometers are the main components of the IMU, the accuracy of which directly affects the accuracy of the inertial system. In general, an IMU includes three single-axis accelerometers and three single-axis gyroscopes, the accelerometers detecting acceleration signals of the object in three independent axes of the carrier coordinate system, and the gyroscopes detecting angular velocity signals of the carrier relative to the navigation coordinate system, measuring the angular velocity and acceleration of the object in three-dimensional space, and calculating the attitude of the object based thereon. Has important application value in navigation.

4. A camera: the camera-based image to be recognized can be captured in real time, typically at a frame rate of 30 frames per second, and the current image is recorded.

5. Visual Inertial Odometer (VIO): a self-positioning scheme is made by the fusion camera and the IMU, and for scenes with non-real-time requirements, algorithms such as smoothing (smooth) are generally used, namely all data are referred to, and projection errors are minimized.

6. Map data: referring to fig. 2, fig. 2 is a schematic diagram of map data in an embodiment of the present application, and as shown in the figure, a plurality of nodes exist on a path, and the nodes are locations of the key GPS.

7. Directed graph: referring to fig. 3, fig. 3 is a schematic diagram of a directed graph in an embodiment of the present application, and as shown in the figure, for example, a connecting edge between node 0 and node 1 in the graph has a direction, which indicates that a vehicle can go from node 0 to node 1, but the vehicle cannot go from node 1 to node 0.

With reference to fig. 4, a method for updating map data in the present application will be described below, where an embodiment of the method for updating map data in the present application includes:

101. the method comprises the steps that a server obtains vehicle data, wherein the vehicle data comprise an image to be identified, a position parameter and a direction parameter;

in this embodiment, the server obtains vehicle data from a large number of vehicles, and therefore the vehicle data includes an image stream (i.e., a plurality of frames of images to be recognized), a position parameter, and a direction parameter reported by each vehicle, where the image stream is obtained by shooting with a camera, and the camera may be disposed on the vehicle, may also be installed on the vehicle as an independent camera, and may also be a camera built in a mobile phone or a tablet computer, which is not limited herein. The location parameters may be determined by a positioning system (e.g., GPS or BDS) and the orientation parameters may be determined by the IMU.

The orientation parameter and the position parameter may be complementary during the positioning process. The position parameters have the advantages of high precision and no error dispersion with time, but the positioning range cannot be covered indoors. The directional parameters have the advantages that the positioning range is the full scene, but the positioning accuracy is not high, so that the directional parameters and the position parameters are complemented, and the positioning reliability can be improved.

102. The server generates anchor points according to the vehicle data;

in this embodiment, the server extracts an image stream (i.e., a plurality of frames of images to be recognized), a position parameter, and a direction parameter reported by each vehicle from the vehicle data, and generates an anchor point according to the image stream (i.e., the plurality of frames of images to be recognized), the position parameter, and the direction parameter, that is, the anchor point aligned with different vehicle trajectories can be formed by recognizing a specific object (e.g., a traffic sign) in the sensor, where the anchor point may include a traffic sign at an intersection and a stationary object such as a tree at the intersection.

Specifically, the image stream (i.e., a plurality of frames of images to be recognized) is acquired by the camera, the direction parameters are acquired by the IMU, and the direction parameters include the three-axis attitude angle (or angular velocity) and the acceleration of the object. Based on this, taking the vehicle data collected by the vehicle as an example, in the VIO process, the position information of the anchor point to be matched relative to the vehicle and the attitude information of the anchor point relative to the vehicle can be acquired in a loose coupling or tight coupling mode. For convenience of understanding, please refer to fig. 5, fig. 5 is a schematic diagram of outputting pose information based on a loose coupling manner in the embodiment of the present application, as shown in the figure, the camera and the IMU are used as two separate modules in the loose coupling, both the two modules can calculate pose information, and then the pose information of the anchor point to be matched is obtained by performing fusion through an Extended Kalman Filter (EKF). Referring to fig. 6, fig. 6 is a schematic diagram of outputting pose information based on a tight coupling mode in the embodiment of the present application, where as shown in the figure, tight coupling refers to a process of processing intermediate data obtained by a camera and an IMU through an optimization filter, and tight coupling requires adding features of an image to be recognized into a feature vector, so as to finally obtain position and posture information of an anchor point to be matched.

The camera and the IMU are fused with good complementarity, the true scale of the camera track can be estimated by aligning the pose sequence estimated by the IMU with the pose sequence estimated by the camera, the IMU can well predict the pose of an image to be recognized and the position of a feature point at the last moment in the next frame of image to be recognized, the matching speed of a feature tracking algorithm and the robustness of the algorithm for responding to rapid rotation are improved, and finally, the gravity vector provided by an accelerometer in the IMU can convert the estimated position into a world coordinate system required by actual navigation.

It should be noted that the anchor points to be matched may be calculated according to vehicle data reported by one vehicle, and in an actual situation, the vehicle data reported by each vehicle needs to be calculated to obtain a plurality of anchor points to be matched, and then the anchor points to be matched are subjected to fusion processing to finally obtain the anchor points on the electronic map.

103. The server generates a fused driving track according to the anchor point and the vehicle data;

in this embodiment, the corresponding anchor points on the electronic map have been generated based on the vehicle data in step 102, and thus accurate positioning of the plurality of vehicles is achieved according to the positions of the anchor points in the world coordinate system (e.g., GPS positions). Because each vehicle has the position parameters at each moment, the corresponding local driving tracks can be generated by combining the position parameters of each vehicle at different moments, and the local driving tracks are superposed and optimized to generate a fused driving track.

104. The server acquires target map data according to the fused driving track, wherein the target map data comprise updated nodes and updated node connecting edges;

in this embodiment, since the fused trajectory generally represents a trajectory curve, after the fused trajectory is obtained, at least two nodes may be extracted from the fused trajectory, where the at least two nodes include a node before turning and a node after turning. In practical applications, in order to improve the positioning accuracy, three or more nodes may be extracted, for example, the nodes before the turn, the nodes in the turn, and the nodes after the turn are included.

Specifically, the electronic map includes target raw map data, which can be regarded as map data corresponding to a route on which a plurality of raw nodes exist, where the plurality of raw nodes at least include nodes before turning and nodes after turning. After the nodes of the fused driving track are extracted, a plurality of updated nodes can be obtained, wherein the plurality of updated nodes at least comprise updated nodes before turning and updated nodes after turning, and the target map data can be formed by connecting the plurality of updated nodes and the nodes.

105. And the server updates the target original map data in the electronic map into target map data.

In this embodiment, the server updates the original target map data to the target map data, that is, the subsequent positioning and navigation are based on the target map data, that is, after the target map data is obtained, the target map data is projected (or projected) back to the original electronic map, that is, the position of the original node in the electronic map is updated. For convenience of description, two scenarios of map data update will be described below as an example.

In an example, please refer to fig. 7, fig. 7 is a schematic diagram illustrating an update of target original map data into target map data according to an embodiment of the present application, where the target original map data is original route data, and taking fig. 7 as an example, the target original map data includes a node a, a node a1, a node B1, a node C1, and a node d. The target map data is the updated route data, and continuing with fig. 7 as an example, the target map data includes node a, node a2, node e, node B2, node f, node C2, and node d. Based on this, updating the node before the turn (updating the node a1 to the node a 2), the node during the turn (updating the node B1 to the node B2) and the node after the turn (updating the node C1 to the node C2), that is, in the case of a delayed turn, the two map data are interchanged, and the update of the map data is realized.

In another example, referring to fig. 8, fig. 8 is another schematic diagram illustrating an update of target original map data into target map data according to an embodiment of the present application, where the target original map data is original route data, and taking fig. 8 as an example, the target original map data includes a node a, a node a1, a node B1, a node C1, and a node B. The target map data is the updated route data, and continuing with fig. 8 as an example, the target map data includes node a, node a2, node B2, node C2, and node B. Based on this, updating the node before the turn (updating the node a1 to the node a 2), the node during the turn (updating the node B1 to the node B2) and the node after the turn (updating the node C1 to the node C2), that is, in the case of a delayed turn, the two map data are interchanged, and the update of the map data is realized.

It should be noted that the above example is merely an illustration, and in practical applications, the map data may be exchanged for a case of turning ahead, so as to update the map data.

Optionally, on the basis of the embodiment corresponding to fig. 3, in another optional embodiment provided in the embodiment of the present application, the obtaining, by the server, the vehicle data may include:

if the vehicle turns before passing through the target node or after passing through the target node, the server receives vehicle data corresponding to the vehicle, wherein the target node is any node included in the target original map data, and the vehicle is one or more vehicles in the N vehicles.

In this embodiment, a method for reporting vehicle data to a server by a terminal device is introduced. Since the positioning system can capture the position parameters of the vehicle, the position parameters include the coordinates of the vehicle in the world coordinate system. And in combination with the position parameters of the vehicle, whether the vehicle data of the vehicle needs to be uploaded to the server or not can be determined.

Specifically, referring to fig. 9 for ease of understanding, fig. 9 is a schematic diagram of an advanced turn or a delayed turn in the embodiment of the present application, and as shown in the figure, the target raw map data includes a node a, a node a2, a node B2, a node C2, and a node B, where the target node is a node a 1. Assuming that the vehicle starts turning before passing through the target node (i.e., the node a 1), the terminal device uploads the vehicle data corresponding to the vehicle to the server, and the driving track of the vehicle passes through the node A3, the node B3, the node C3 and the node B. Assuming that the vehicle starts turning after passing through the target node (i.e., the node a 1), the terminal device uploads the vehicle data corresponding to the vehicle to the server, and the driving track of the vehicle passes through the node a2, the node B2, the node C2 and the node B.

In general, the map data may not meet the actual situation due to road modification or lane change, that is, the original straight road may be curved, or the position of the u-turn may be advanced, or the position of the u-turn may be delayed. These locations are some of the key locations for car navigation, requiring real-time updating of data.

Secondly, in the embodiment of the present application, a method for reporting vehicle data to a server by a terminal device is provided, and by the above method, for a vehicle, a position of the vehicle can be determined in real time based on a positioning system, so that whether the vehicle turns ahead at a place where the vehicle needs to turn or turns after a delay can be detected, and thus whether the vehicle data of the vehicle is uploaded to the server is determined. If the vehicle does not turn ahead or turn behind, the vehicle data does not need to be reported to the server, so that resources consumed by data transmission are saved, if the vehicle turns ahead or turns behind, the vehicle data is automatically reported so that the server can update the map data based on the data, when enough vehicles pass through the same intersection, the map data can be updated according to the driving track, and therefore more accurate map data can be obtained.

Optionally, on the basis of the embodiment corresponding to fig. 3, in another optional embodiment provided by the embodiment of the present application, the generating, by the server, the anchor point according to the vehicle data may include:

the server acquires an interested area according to the image to be identified included in the vehicle data;

the server acquires the characteristic points corresponding to the region of interest;

the server generates extended characteristic information according to the characteristic points corresponding to the region of interest and the direction parameters and the position parameters included in the vehicle data;

the server generates three-dimensional extended characteristic information corresponding to the region of interest according to the characteristic points corresponding to the region of interest and the extended characteristic information;

and the server determines the position corresponding to the anchor point in the electronic map according to the three-dimensional expansion characteristic information corresponding to the region of interest.

In this embodiment, a manner of generating an anchor point based on vehicle data is described. The server inputs the image to be recognized into the trained image recognition model, the region of interest is output through the image recognition model, then feature points are extracted from the region of interest, and the feature points are usually represented by 128-dimensional feature descriptors with unchanged illumination and consistent dimensions. Further, the server may generate a relative angle between the vehicle and the region of interest, expressed as a 3-dimensional feature, from the direction parameters. The server may also generate a relative position between the vehicle and the region of interest from the position parameters, the relative position being represented as a 3-dimensional feature. Based on this, the 128-dimensional feature descriptors, the 3-dimensional relative angle and the 3-dimensional relative position are combined to obtain 134-dimensional extended feature information, wherein the extended feature information is embodied in the form of extended feature descriptors.

It should be noted that the manner of Feature point extraction includes, but is not limited to, extracting Feature points based on Scale-Invariant Feature Transform (SIFT), testing Feature operators and rotation based on Accelerated Robust Features (SURF), and Binary Robust Independent basis Feature (ORB) algorithm based on Oriented Accelerated segmentation and Feature segmentation.

Specifically, the server combines the three-dimensional coordinates of the feature point corresponding to the region of interest with the extended feature information, thereby obtaining three-dimensional extended feature information corresponding to the region of interest, where the three-dimensional extended feature information may be represented as a 137-dimensional feature, where the three-dimensional extended feature information is embodied in the form of a three-dimensional extended feature descriptor. Based on the three-dimensional extended feature information corresponding to each region of interest, the real coordinates of the object included in the regions of interest in the world coordinate system can be respectively calculated, and then the position corresponding to the anchor point is determined based on the type of the object.

Secondly, in the embodiment of the application, a mode for generating anchor points based on vehicle data is provided, and through the mode, the positions of the anchor points under a world coordinate system can be fitted according to vehicle data from a large number of vehicles, so that the situation that the position deviation of the anchor points is large due to small quantity or inaccurate positioning data is avoided, and the accuracy of the anchor point positions is improved.

Optionally, on the basis of the embodiment corresponding to fig. 3, in another optional embodiment provided by the embodiment of the present application, acquiring the region of interest according to the image to be identified included in the vehicle data may include:

acquiring an image recognition result through an image recognition model based on an image to be recognized included in the vehicle data;

In this embodiment, a manner of outputting a region of interest by using an image recognition model is described. Firstly, an image recognition model needs to be obtained through pre-training, the image recognition model may adopt a Convolutional Neural Networks (CNN) structure, or may adopt a network structure such as the Visual Geometry Group (VGG) of the oxford university, which is not limited herein. Then, inputting the acquired to-be-recognized images of each frame into a trained image recognition model, and outputting corresponding image recognition results by the image recognition model, wherein the image recognition results can comprise segmentation region frames corresponding to the region of interest, object classes and class probabilities.

Specifically, the server may further determine whether an indication mark or a stationary reference object exists according to the image recognition result. Assuming that the image recognition result includes a region of interest a, a region of interest B, a region of interest C, a region of interest D, and a region of interest E, for convenience of description, please refer to table 1, where table 1 is an illustration of the image recognition result.

TABLE 1

Assuming that the recognition probability threshold is 50%, as can be seen from table 1, the segmented region box 1 extracts the region of interest a, and it is determined that the object in the region of interest a may be a traffic sign after recognition, and the probability of belonging to the traffic sign is 60% (i.e., greater than the recognition probability threshold), thereby determining that the indication mark exists in the region of interest a. The region B extracted from the segmentation region box 2 is identified to determine that the object in the region B may be a tree and the probability of belonging to the tree is 77% (i.e. greater than the identification probability threshold), thereby determining that a stationary reference object exists in the region B. The region C extracted from the segmentation area block 3 is identified to determine that the object in the region C may be a ground mark and the probability of belonging to the ground mark is 39% (i.e. less than the identification probability threshold), thereby determining that no indication mark exists in the region C. The region of interest D extracted by the segmentation region block 4 is identified to determine that the object in the region of interest D is likely to be a stone and has a probability of belonging to the stone of 28% (i.e., less than the identification probability threshold), thereby determining that no stationary reference object exists in the region of interest D. The region E extracted by the segmentation region frame 5 is a region of interest, and it is determined that the object in the region of interest E is possibly a pedestrian after recognition, and the pedestrian does not belong to the indication mark nor the static reference object.

It should be noted that the indication signs include, but are not limited to, text traffic signs, graphic traffic signs, and ground signs. The word traffic board is mainly a traffic board for transmitting specific information through words, the graphic traffic board is mainly a traffic board for transmitting specific information through graphic symbols, the ground mark is mainly a mark object arranged on the ground, for example, at two ends of a one-way road, a direction allowing traffic is written with a one-way driving direction, a large arrow for forward driving is marked, and a direction forbidding traffic of a motor vehicle is marked in a direction forbidding traffic, and meanwhile, a forbidding traffic mark is also marked. The stationary reference object includes, but is not limited to, a stationary type object or scene such as a non-human and non-vehicle.

In practical applications, the feature points should be distributed as much as possible within the main region of interest, and there may be a small number of feature points within the sub-region of interest, but it should be noted that, in the calculation process, feature points should be set in both the main region of interest and the sub-region of interest.

In the embodiment of the application, a mode of outputting the region of interest by using the image recognition model is provided, and by the above mode, the trained image recognition model can be used for quickly and accurately extracting the main region of interest and the secondary region of interest in the image to be recognized, and further recognizing the type of the object in the region of interest, so that the accuracy of anchor point selection can be improved, and the reliability of positioning can be improved.

Optionally, on the basis of the embodiment corresponding to fig. 3, in another optional embodiment provided by the embodiment of the present application, the determining, by the server, the position corresponding to the anchor point in the electronic map according to the three-dimensional extended feature information corresponding to the area of interest may include:

the server determines the three-dimensional expansion characteristic information corresponding to the interested area as the three-dimensional expansion characteristic information corresponding to the anchor point to be matched;

the server acquires three-dimensional expansion characteristic information corresponding to the first candidate anchor point;

the server matches the three-dimensional extended feature information corresponding to the anchor point to be matched with the three-dimensional extended feature information corresponding to the first candidate anchor point;

if the three-dimensional expansion feature information corresponding to the anchor point to be matched is successfully matched with the three-dimensional expansion feature information corresponding to the first candidate anchor point, the server fuses the anchor point to be matched and the first candidate anchor point to obtain the position corresponding to the anchor point in the electronic map;

and if the three-dimensional expansion feature information corresponding to the anchor point to be matched fails to be matched with the three-dimensional expansion feature information corresponding to the first candidate anchor point, the server determines the anchor point to be matched as a second candidate anchor point.

In this embodiment, a method of adding a candidate anchor point based on incremental matching is introduced. Based on the vehicle data and the three-dimensional expansion feature information corresponding to the region of interest, the coordinates of the anchor point to be matched under the world coordinate system can be converted, wherein the anchor point to be matched is a certain object in the region of interest, such as a graphical traffic sign. Based on the above, the three-dimensional extended feature information corresponding to the anchor point to be matched is matched with the three-dimensional extended feature information corresponding to the first candidate anchor point.

Specifically, descriptors (for example, 128-dimensional descriptors) corresponding to the feature points are extracted from three-dimensional extended feature information corresponding to the anchor points to be matched, similarly, descriptors (for example, 128-dimensional descriptors) corresponding to the feature points are extracted from three-dimensional extended feature information corresponding to the first candidate anchor points, then the similarity between the two descriptors can be calculated, and if the similarity is greater than a threshold value, the matching is successful, so that the anchor points to be matched, the first candidate anchor points and other similar anchor points can be fused to obtain the final anchor point position. If the similarity is smaller than the threshold, the matching is failed, and then the anchor point to be matched is determined as a second candidate anchor point, and the second candidate anchor point can be used for subsequent anchor point matching.

And thirdly, in the embodiment of the application, a mode of adding the candidate anchor point based on the incremental matching is provided, through the mode, after the candidate anchor point is obtained, whether the candidate anchor point is successfully matched with the existing candidate anchor point needs to be judged, if the matching is successful, the candidate anchor point belongs to the real anchor point, and if the matching is failed, the candidate anchor point is used as a new candidate anchor point for subsequent matching. Therefore, candidate anchor points are added in an incremental matching mode, on one hand, more accurate anchor points can be selected, on the other hand, full matching is not needed each time, and resource consumption in the matching process can be reduced.

Optionally, on the basis of the embodiment corresponding to fig. 3, in another optional embodiment provided by the embodiment of the present application, the matching, by the server, the three-dimensional extended feature information corresponding to the anchor point to be matched and the three-dimensional extended feature information corresponding to the first candidate anchor point may include:

the server determines the matching degree according to the three-dimensional expansion characteristic information corresponding to the anchor point to be matched and the three-dimensional expansion characteristic information corresponding to the first candidate anchor point;

after the server matches the three-dimensional extended feature information corresponding to the anchor point to be matched with the three-dimensional extended feature information corresponding to the first candidate anchor point, the method further comprises the following steps:

if the matching degree is greater than or equal to the threshold value of the matching degree, the server determines that the three-dimensional expansion feature information corresponding to the anchor point to be matched is successfully matched with the three-dimensional expansion feature information corresponding to the first candidate anchor point;

and if the matching degree is less than or equal to the threshold value of the matching degree, filtering the anchor points to be matched by the server.

In this embodiment, a method of filtering anchors to be matched in the matching process is introduced. Since there may be errors in the process of collecting vehicle data, and these error data may have a certain influence on the calculation result, it is also necessary to reject these data with larger errors, and a processing manner will be described below.

Specifically, a relative position is extracted from three-dimensional extended feature information corresponding to an anchor point to be matched, similarly, a relative position is extracted from three-dimensional extended feature information corresponding to a first candidate anchor point, a matching degree between the two is calculated based on the relative positions of the two, and a relation between the matching degree and a matching degree threshold is judged, so that whether the anchor point to be matched is filtered or not is determined.

Exemplarily, it is assumed that 100 vehicles pass through an anchor point a to be matched, a candidate anchor point B, a candidate anchor point C and a candidate anchor point D, for the anchor point a to be matched, there is a cosine similarity between the 1 st vehicle and the 2 nd vehicle, there is a cosine similarity between the 2 nd vehicle and the 3 rd vehicle, and so on, there is a cosine similarity between the 99 th vehicle and the 100 th vehicle, and the cosine similarities of two vehicles in the 100 vehicles are averaged to obtain a matching degree threshold. Based on this, the matching degree (e.g., cosine similarity) is determined according to the relative position corresponding to the anchor point to be matched (e.g., anchor point a to be matched) and the relative position corresponding to the first candidate anchor point (e.g., candidate anchor point B). And if the matching degree is greater than or equal to the threshold value of the matching degree, determining that the three-dimensional extended feature information corresponding to the anchor point to be matched is successfully matched with the three-dimensional extended feature information corresponding to the first candidate anchor point. Otherwise, if the matching degree is smaller than or equal to the threshold value of the matching degree, filtering the anchor points to be matched.

For example, feature point descriptors corresponding to a plurality of three-dimensional extended feature information may be clustered, so as to determine whether the anchor point to be matched needs to be filtered, and if the anchor point to be matched is far from the central point, the anchor point to be matched may be filtered.

It should be noted that, in practical applications, the anchor point to be matched with a large error may also be filtered by using the covariance of the positioning, for example, if the positioning of the vehicle a is not accurate (for example, a failure of an antenna for transmitting and receiving signals is detected), then the confidence of the position parameter acquired by the vehicle a is low, the quality of the obtained data is also poor, and the covariance of the positioning is also large, so that the anchor point to be matched determined based on the vehicle data provided by the vehicle a may be filtered.

It should be noted that, in practical applications, the location parameter at a certain specific location may also be filtered, for example, there is a shelter at a certain specific location, which may affect the transmission and reception of signals, so that the confidence of all the location parameters collected through the tree is low, the quality of the obtained data is also poor, and the filtering may be based on the anchor point to be matched determined when passing through the tree.

Further, in the embodiment of the present application, a manner of filtering anchors to be matched in a matching process is provided, and in the manner, since vehicle data includes data reported from a large number of vehicles, it is suggested that there may be some data with large errors due to differences of devices, and the data belongs to "noise" in a data processing process, and in order to improve accuracy of anchor positions, the anchors to be matched with large errors need to be filtered, so that reliability of anchor calculation is improved.

Optionally, on the basis of the embodiment corresponding to fig. 3, in another optional embodiment provided by the embodiment of the present application, the generating, by the server, the fused trajectory according to the anchor point and the vehicle data may include:

the server generates multidimensional Gaussian distribution corresponding to the electronic map according to the vehicle data;

the server generates a two-dimensional driving probability map corresponding to the electronic map according to the anchor points and the vehicle data;

the method comprises the steps that a server determines lane types according to multi-dimensional Gaussian distribution, wherein the lane types are single lane types or multi-lane types;

and the server generates a fused driving track according to the lane type and the two-dimensional driving probability map.

In this embodiment, a method for generating a fused driving track based on vehicle data is introduced. The server generates multidimensional Gaussian distribution corresponding to the electronic map according to the vehicle data, and determines that the vehicle type is a single lane type or a multi-lane type according to the condition of the multidimensional Gaussian distribution.

In the process of generating the two-dimensional driving probability map, the server needs to optimize each local driving track. Specifically, the position of the anchor point is fixed, and each local trajectory is optimized by using a Bundle Adjustment (BA) algorithm, so as to obtain an optimized local trajectory, wherein the optimized local trajectory belongs to a trajectory under world coordinates. The position of an anchor point in a scene is fixed by a beam adjustment algorithm, then the position of a vehicle is optimized, light rays which are emitted from the optical center of a camera corresponding to each view and pass through pixels corresponding to the anchor point in an image are intersected with the anchor point, and a great number of light beams (bundle) are formed for all three-dimensional points. In an actual process, because noise and the like exist, each ray is almost impossible to converge with one point, and therefore, in a solving process, the required information needs to be continuously adjusted (adjustment) so that the final ray can be handed over to the anchor point.

Secondly, in the embodiment of the application, a mode of generating the fused driving track based on the vehicle data is provided, the anchor point is fixed through the mode, then the node position on the driving track is adjusted through a light beam method adjustment algorithm, and finally a more accurate fused driving track is obtained, so that the feasibility and the reliability of the scheme are improved.

Optionally, on the basis of the embodiment corresponding to fig. 3, in another optional embodiment provided by the embodiment of the present application, the generating, by the server, the two-dimensional travel probability map corresponding to the electronic map according to the anchor point and the vehicle data may include:

the server constructs a two-dimensional original map corresponding to the electronic map, wherein the two-dimensional original map comprises (N multiplied by M) grids, and both N and M are integers greater than 1;

the server acquires K local driving tracks according to the position parameters included in the vehicle data, wherein the vehicle data include vehicle data of K vehicles, the vehicle data of each vehicle are used for generating one local driving track, and K is an integer larger than 1;

the server performs optimization processing on each local trajectory in the K local trajectories according to the anchor points to obtain K optimized local trajectories;

and the server maps the K optimized local driving tracks to a two-dimensional original graph to obtain a two-dimensional driving probability graph, wherein the probability value corresponding to the grid in the two-dimensional driving probability graph is in positive correlation with the passing times of the K optimized local driving tracks.

In the present embodiment, a manner of generating a two-dimensional travel probability map from vehicle data is described. The server can generate a corresponding two-dimensional original map from the electronic map, where the two-dimensional original map can be understood as a grid map composed of (N × M) grids, where N and M are integers greater than 1, each grid corresponds to one pixel, and each pixel represents a real size of 0.5 × 0.5 meters. After the two-dimensional original map is generated, corresponding local driving tracks are respectively generated based on the position parameters of each vehicle in the vehicle data at different moments, and K local driving tracks are obtained on the assumption that the position parameters of K vehicles in the vehicle data at different moments are total, wherein K is an integer greater than 1. Based on the foregoing embodiment, it can be known that each local trajectory can be optimized based on the anchor point that has been fixed, and the optimization manner is as described in the foregoing embodiment, which is not described herein again.

Specifically, each grid in the two-dimensional original map has an occupancy rate, and for each grid, the occupancy rate is represented as "1" if the grid is fully occupied, i.e., the corresponding probability value is "1", and the occupancy rate is represented as "0" if the grid is not fully occupied, i.e., the corresponding probability value is "0". Thus, values in 0 to 1 may also be used in a grid to indicate that the grid is occupied. In one case, assuming a total of 100 vehicles, when a vehicle passes through a grid, the probability value of the grid is added to 0.01, assuming 60 vehicles pass through the grid, the probability value of the grid is 0.6, and assuming 100 vehicles pass through the grid, the probability of the grid is 1. In another case, when a vehicle passes through a grid, the probability value of the grid is increased by 0.001, the probability value of the grid is 0.1 assuming that 1000 vehicles pass through the grid, and the probability value of the grid is 1 assuming that 50000 vehicles pass through the grid, regardless of the number of vehicles.

It should be noted that, in practical applications, the calculation strategy of the grid probability value may also be adjusted according to actual requirements, and the above examples are only two illustrations and should not be construed as limiting the present application.

For convenience of understanding, please refer to fig. 11, fig. 11 is a schematic diagram of a two-dimensional original map and a two-dimensional driving probability map in an embodiment of the present application, as shown in fig. 11 (a), the two-dimensional original map includes 24 × 14 grids, and a probability value of each grid in an initial state is 0. After the K optimized local trajectories are mapped to the two-dimensional original map, a two-dimensional driving probability map as shown in (B) of fig. 11 is obtained, and it can be seen on the two-dimensional driving probability map that some grids are white, some grids are gray, and some grids are black, where a pixel point corresponding to the white grid indicates that no vehicle passes through, a pixel point corresponding to the gray grid indicates that a part of vehicles pass through, and a pixel point corresponding to the black grid indicates that a large number of vehicles pass through. Therefore, the probability values corresponding to the grids in the two-dimensional driving probability map are in positive correlation with the passing times of the K optimized local driving tracks, that is, for one grid, the more the K optimized local driving tracks pass, the greater the probability value of the grid is. Based on the method, the K optimized local driving tracks are fused into the two-dimensional driving probability map.

In the embodiment of the application, a mode for generating a two-dimensional driving probability map according to vehicle data is provided, through the above mode, position parameters of different vehicles at different moments are extracted from the vehicle data reported by a large number of vehicles, then, according to the position parameters of each vehicle at different moments, the optimized local driving track is recorded in a two-dimensional original map, and finally, according to the local driving tracks of all the vehicles on the two-dimensional original map, a corresponding two-dimensional driving probability map is generated, so that the effect of fusing the optimized local driving tracks is achieved, and the authenticity and the reliability of the two-dimensional driving probability map are improved.

Optionally, on the basis of the embodiment corresponding to fig. 3, in another optional embodiment provided by the embodiment of the present application, the determining, by the server, the lane type according to the multidimensional gaussian distribution may include:

the server acquires the length of the primary direction and the length of the secondary direction according to the multi-dimensional Gaussian distribution;

the server generates a flat parameter according to the length in the primary direction and the length in the secondary direction;

the server determines a segmentation threshold value according to the flat parameters;

the server determines the lane type based on the segmentation threshold.

In the present embodiment, a method for determining a vehicle type based on a multidimensional gaussian distribution is described. According to the embodiment, the server generates the multidimensional Gaussian distribution corresponding to the electronic map according to the vehicle data, selects the appropriate segmentation threshold according to the multidimensional Gaussian distribution, and determines the corresponding lane type by combining the segmentation threshold.

Specifically, a plane formed by X and Y in the multidimensional gaussian distribution can be regarded as a plane corresponding to the electronic map, and the Z-axis direction represents the probability density. For easy understanding, please refer to fig. 10, fig. 10 is a schematic diagram of a relationship between a lane type and a multidimensional gaussian distribution in the embodiment of the present application, as shown in (a) of fig. 10, after the multidimensional gaussian distribution is mapped to a plane, if a distinct peak can be seen, the lane type is identified as a single lane. After mapping the multidimensional gaussian distribution to a plane, as shown in fig. 10 (B), if one can be seen to be relatively flat, it is recognized that the lane type is multilane.

For convenience of understanding, referring to fig. 12, fig. 12 is a schematic diagram of a multi-dimensional gaussian distribution in the embodiment of the present application, as shown in (a) of fig. 12, since the gaussian distribution adopted in the present application is a multi-dimensional gaussian distribution, there are values in the X-axis direction, the Y-axis direction and the Z-axis direction. Taking the multi-dimensional gaussian distribution as a two-dimensional gaussian distribution as an example, if two dimensions of random variables are uncorrelated and the covariance matrix is a diagonal matrix, the top view of the two-dimensional gaussian distribution is shown in fig. 12 (B), and thus it can be seen that the top view of the two-dimensional gaussian distribution is circular. If two dimensions of the random variables are correlated and the covariance matrix is a symmetric matrix, the top view of the two-dimensional gaussian distribution is shown in fig. 12 (C), and thus it can be seen that the top view of the two-dimensional gaussian distribution is elliptical.

Based on this, a point with the highest probability density (i.e. the point with the highest heat) can be selected from the multi-dimensional gaussian distribution, the point is determined as the maximum value, the maximum value is used as the center point to circle a range, for convenience of description, please refer to fig. 13, fig. 13 is a schematic diagram of the length in the primary direction and the length in the secondary direction in the embodiment of the present application, as shown in the figure, a range is circled after the maximum value is used as the center point, the range is represented as an ellipse, and it can be seen that the larger the value on the X axis is, the smaller the value on the Y axis is, therefore, this situation can be referred to. As shown in fig. 13, a1 indicates the minor direction length, and a2 indicates the major direction length, and the flatness parameter is calculated as follows:

S=length1/length2；

where S denotes a flatness parameter, length1 denotes a minor direction length, and length2 denotes a major direction length. It can be seen that the larger the flattening parameter, the more circular the shape of the ellipse is represented, and the smaller the flattening parameter, the flatter the shape of the ellipse is represented.

Next, a corresponding segmentation threshold may be determined according to the flat parameter, please refer to table 2, where table 2 is an illustration of a corresponding relationship between the flat parameter range and the segmentation threshold.

TABLE 2

Taking table 2 as an example, assuming that the flat parameter is 0.75, it is determined that the traffic lane falls within the (0.7, 0.8) interval, and therefore, the corresponding segmentation threshold is 50%.

In the embodiment of the application, a mode for determining the type of the vehicle based on the multidimensional gaussian distribution is provided, and through the mode, the corresponding flat parameters can be generated by combining the specific situation of the multidimensional gaussian distribution, and the segmentation threshold is determined based on the flat parameters, so that a more reasonable segmentation threshold can be generated, and the purpose of dynamically adjusting the segmentation threshold is achieved.

Optionally, on the basis of the embodiment corresponding to fig. 3, in another optional embodiment provided by the embodiment of the present application, the determining, by the server, the lane type according to the segmentation threshold may include:

the server performs binarization processing on the two-dimensional driving probability map by adopting a segmentation threshold value to obtain a binarization driving probability map, wherein the binarization driving probability map comprises a first region, and the probability value of each pixel point in the first region is greater than or equal to the segmentation threshold value;

the server acquires a second region according to the two-dimensional driving probability map, wherein the probability value of each pixel point in the second region is greater than 0;

the server determines an intersection ratio according to the first area and the second area;

if the intersection ratio is smaller than or equal to the intersection ratio threshold value, the server determines that the lane type is a single lane type;

and if the intersection ratio is greater than the intersection ratio threshold value, the server determines that the lane type is a multi-lane type.

In the present embodiment, a manner of determining the type of the vehicle based on the segmentation threshold is described. As can be seen from the foregoing embodiment, after the segmentation threshold is obtained, binarization processing may be performed on the two-dimensional driving probability map to obtain a binarization driving probability map, where the binarization driving probability map is a probability map composed of a grid with a pixel probability value of "1" and a grid with a pixel probability value of "0", and the grids with the probability values of "1" are used together as the first region. In addition, a second region is further acquired according to the two-dimensional driving probability map, the second region includes at least one grid, the probability value of a pixel point corresponding to each grid is greater than 0, and based on the probability value, an Intersection Over Unit (IOU) between the first region and the second region can be calculated.

It will be appreciated that the intersection-to-union ratio is a criterion that measures the accuracy of detecting the corresponding object in a particular data set, and is calculated as the ratio of the intersection and union of the "predicted bounding box" and the "true bounding box". And if the intersection ratio between the first area and the second area is less than or equal to the intersection ratio threshold, determining that the lane type is a single lane type. Determining the lane type as a multi-lane type if the intersection ratio between the first area and the second area is greater than the intersection ratio threshold. The threshold value of the cross-over ratio may be set to 20% or 30% in a general case, and may also be set according to actual circumstances.

For convenience of understanding, please refer to fig. 14, fig. 14 is a schematic diagram of the intersection ratio between the first region and the second region in the embodiment of the present application, as shown in fig. 14 (a), there are white, gray and black grids in the two-dimensional driving probability map, and if the pixel point probability value corresponding to the gray grid is 60% and the segmentation threshold value is 80%, the binarized driving probability map obtained after the binarization processing is performed on the two-dimensional driving probability map is as shown in fig. 14 (B), so that only black and white grids exist in the binarized driving probability map. The binary driving probability map includes a first region, wherein the first region includes 54 black grids. A second region is included in the two-dimensional driving probability map, wherein the second region includes 75 gray grids and 54 black grids. Thus, an intersection ratio of 54/(54+75) =0.42, that is, 42% is obtained between the first region and the second region. Assuming that the intersection ratio threshold is 30%, it may be determined that the lane type is a multi-lane type.

Further, in the embodiment of the present application, a manner of determining a vehicle type based on a segmentation threshold is provided, and in the manner, after the segmentation threshold is determined, binarization processing may be performed on the two-dimensional driving probability map based on the segmentation threshold, so that all pixel points greater than or equal to the segmentation threshold are selected, and then a second region is obtained in the two-dimensional driving probability map, so that a merging ratio may be determined, and a lane type may be determined more accurately based on the merging ratio, thereby improving feasibility and operability of a scheme.

Optionally, on the basis of the embodiment corresponding to fig. 3, in another optional embodiment provided by the embodiment of the present application, the generating, by the server, the fused vehicle trajectory according to the lane type and the two-dimensional driving probability map may include:

if the lane type is a single lane type, the server performs curve fitting processing on a first area included in the binary driving probability map to generate a fused driving track;

the server acquires the target map data according to the fused driving track, and the method may include:

and the server extracts Q updated nodes from the fused driving track, wherein Q is an integer larger than 1, and the Q updated nodes at least comprise nodes before turning and nodes after turning.

In this embodiment, a method for generating a fused driving track based on a single lane is introduced. As can be seen from the description in the foregoing embodiment, the server generates the segmentation threshold according to the flat parameter, then performs binarization processing on the two-dimensional driving probability map by using the segmentation threshold, that is, determines the probability value of each pixel, sets the probability value to "1" for pixels having a probability value greater than or equal to the segmentation threshold, and sets the probability value to "0" for pixels having a probability value less than the segmentation threshold, and finally generates a binarization driving probability map with only "1" and "0". And extracting all pixel points with the probability value of 1 in the binarization driving probability map to obtain a first region.

Specifically, if the lane type is a single lane type, after the first region is obtained, the first region may be directly extracted from the binarized driving probability map, and then curve fitting processing may be performed on the first region to generate the fused vehicle trajectory. It should be noted that, first, each pixel point is taken as a discrete data point, then, one or more of the following ways of curve fitting can be adopted, and the curve fitting ways include but are not limited to curve fitting by adopting a straight line fitting regression equation, curve fitting by adopting a quadratic polynomial fitting regression equation, curve fitting by adopting a cubic polynomial fitting regression equation, curve fitting by adopting a semilogarithmic fitting regression equation, curve fitting by adopting a Log-Log fitting regression equation, curve fitting by adopting a four-parameter fitting regression equation, curve fitting by adopting cubic spline interpolation, and curve fitting by adopting point-to-point. In addition, curve fitting can also be realized in the form of a piecewise function.

Taking turning as an example, the server extracts Q updated nodes based on the angle of the fused driving track, where Q is an integer greater than 1, that is, extracts at least two updated nodes. The at least two updated nodes here need to include a node before the turn that can identify a start position where the vehicle starts turning and a node after the turn that can identify an end position where the vehicle completes turning. In practical applications, one or more nodes in the turning process can also be extracted, for example, a point in the middle of the fused driving track is determined as a node in the turning process.

Thirdly, in the embodiment of the application, a method for generating a fused traffic track based on a single lane is provided, through the method, a segmentation threshold value is adopted to carry out binarization processing on a two-dimensional traffic probability map to obtain a binarization traffic probability map, under the condition that the lane type is determined to be the single lane type, curve fitting processing is carried out on a first area included in the binarization traffic probability map to generate the fused traffic track, and the fused traffic track is optimized based on a beam adjustment algorithm to obtain the fused traffic track, so that at least two updated nodes can be extracted from the fused traffic track, and the feasibility and the operability of the scheme are improved.

if the lane type is a single lane type, the server constructs a polygonal area according to a first area included in the binary driving probability map;

the server generates a fused driving track according to the polygonal area;

In this embodiment, a method for generating a fused driving track based on multiple lanes is introduced. As can be seen from the description in the foregoing embodiment, the server generates the segmentation threshold according to the flat parameter, then performs binarization processing on the two-dimensional driving probability map by using the segmentation threshold, that is, determines the probability value of each pixel, sets the probability value to "1" for pixels having a probability value greater than or equal to the segmentation threshold, and sets the probability value to "0" for pixels having a probability value less than the segmentation threshold, and finally generates a binarization driving probability map with only "1" and "0". And extracting all pixel points with the probability value of 1 in the binarization driving probability map to obtain a first region.

Specifically, if the lane type is a multi-lane type, after the first region is obtained, a polygonal region may be first constructed based on the first region, for convenience of understanding, refer to fig. 15, where fig. 15 is a schematic diagram of extracting a fused driving track from the polygonal region in the embodiment of the present application, as shown in the figure, external contours are directly extracted for multiple lanes, and then bones are further extracted, where B1 indicates the external contours, and B2 indicates the bones based on the external contour extraction process. In one case, the skeleton is directly used as the fusion trajectory. In another case, a curve fit may be performed on the skeleton to generate a fused vehicle trajectory.

It should be noted that a skeleton is an important geometric feature of an image, and the skeleton of an image is to thin a polygonal region having a certain area into one line. The skeleton is understood to mean the central axis of the image, for example, a rectangular skeleton, which is the central axis in its longitudinal direction, and for example, a circular skeleton is the center of the circle, a straight skeleton is itself, and a skeleton of an isolated point is itself. The skeleton is obtained mainly by two methods, the first method is based on the simulation of the burning fire, the process is similar to that the edge lines of the external outline are all ignited at the same moment, the front edge of the fire spreads inwards at a constant speed, the flame is extinguished when the front edges are intersected, and the combination of the flame extinguishing points is the skeleton. The second is based on the largest disc, that is, the skeleton of the polygon region is defined as the set of all largest inscribed circles inside the polygon region. If a disk with the diameter capable of being changed at will is provided, the disk is placed in the polygonal area, the diameter of the disk is changed continuously, all the maximum inscribed circles which are contained in the polygonal area and have at least two tangent points with the boundary of the polygonal area are found out, and the track formed by the circles of all the maximum inscribed circles is the framework of the polygonal area.

The embodiment of the application provides a method for generating a fused vehicle trajectory based on multiple lanes, and through the method, a segmentation threshold value is adopted to carry out binarization processing on a two-dimensional driving probability map to obtain a binarized driving probability map, under the condition that the lane type is determined to be the multiple lane type, a polygonal area is formed according to a first area included in the binarized driving probability map, then a skeleton in the polygonal area is extracted, so that a fitted fused vehicle trajectory is obtained, and the fused vehicle trajectory is optimized based on a light beam method adjustment algorithm to obtain the fused vehicle trajectory, so that at least two updated nodes can be extracted from the fused vehicle trajectory, and therefore feasibility and operability of the scheme are improved.

Referring to fig. 16, please refer to fig. 16, wherein fig. 16 is a schematic overall flowchart of a map data updating method in an embodiment of the present application, and as shown in the figure, specifically:

in step S1, the image recognition model extracts regions of interest, where the regions of interest include a main region of interest and a sub-region of interest, and the main region of interest mainly includes a traffic sign, a graphic traffic sign, a ground sign, and the like. The secondary region of interest mainly includes other non-human and non-vehicular stationary types of scene regions. The extraction of the regions of interest is prepared for the next step of extracting the feature points, so that the relatively clean background feature points can be obtained.

In step S2, feature point extraction is performed on the region of interest to obtain a descriptor, and then the relative position and the relative angle of the vehicle are combined to obtain an extended feature descriptor, where the feature point extraction method may be an ORB method or an SIFT method, and generally uses a light whiteboard and feature descriptors with uniform scale.

In step S3, the feature points are triangulated to extract three-dimensional coordinates, and a three-dimensional extended feature descriptor is obtained together with the extended descriptor. The global positioning coordinates (e.g., GPS coordinates) of the image points may be calculated, in particular, based on vehicle data extracted by the camera, IMU, and a positioning system (e.g., GPS), in conjunction with a smoothing algorithm. Assuming that all junctions are horizontal, three-dimensional free (3 DoF) modeling, i.e., planar XY coordinates and yaw (yaw), is employed.

In step S4, a global plurality of pairs is obtained by using an incremental matching method according to different vehicle data or data of different time periods. The incremental matching means that the three-dimensional extended feature descriptors generated by each vehicle are matched into a global space, the matching is successful and added into the existing matching pair, and the matching is failed and added into a new matching candidate pair.

In step S5, based on the different pair-fused coordinates, globally unique coordinates are obtained, that is, assuming that the coordinates obey gaussian distribution, all data in each pair are fused based on the gaussian fusion method.

In step S6, some obvious outliers (outlers) are filtered to obtain a plurality of anchor points, wherein the filtering rule may use the covariance of the GPS, the matching degree of the features, and the like.

In step S7, the anchor point position is fixed, and the map optimizes each segment of the driving track and the gesture, and the vehicle data extracted by the camera, the IMU and the positioning system (e.g. GPS) is essentially calculated by using smooth algorithm, and the difference is mainly that the anchor point position is fixed, i.e. the anchor point position is not optimized, and a strong track constraint is provided. Assuming that all junctions are horizontal, 3DoF modeling, i.e., planar XY coordinates and yaw (yaw), is used.

In step S8, the travel probability map is binarized by a certain division threshold value, and is divided into a single lane or a plurality of lanes according to the intersection ratio region. The method comprises the steps of carrying out binarization processing on a two-dimensional driving probability map by using a certain segmentation threshold value, and then distinguishing whether the lane type is a single lane or a multi-lane based on intersection and comparison.

In step S9, at least two nodes are extracted according to the division of the lane type. Taking three nodes as an example, assuming that the lane type is a single lane, directly fitting a curve to obtain a fused driving track, and extracting a node A, a node B and a node C according to the turning type, wherein the node A is a node before turning, the node B is a node during turning, and the node C is a node after turning.

Assuming that the lane type is a multi-lane, performing threshold segmentation on the two-dimensional driving probability map to obtain a binary driving probability map, filling up a void in the binary driving probability map, searching for an edge contour, obtaining a polygonal area after the edge contour is protruded, extracting a skeleton of the polygonal area as a fusion driving track, and extracting a node A, a node B and a node C according to a single-lane processing mode.

In step S10, the map data in the electronic map is updated according to the Identification (ID) value of each node, and since the map data in the electronic map is stored in a graph structure, each node has a corresponding ID value, and the corresponding node is found according to the ID value, and updating or error correction can be performed.

Referring to fig. 17, fig. 17 is a schematic diagram of an embodiment of a map data updating apparatus 20 according to the present application, which includes:

the acquiring module 201 is configured to acquire vehicle data, where the vehicle data includes an image to be identified, a position parameter, and a direction parameter;

a generating module 202, configured to generate an anchor point according to the vehicle data;

the generating module 202 is further configured to generate a fused driving track according to the anchor point and the vehicle data;

the obtaining module 201 is further configured to obtain target map data according to the fused driving track, where the target map data includes updated nodes and updated node connecting edges;

and the updating module 203 is used for updating the target original map data in the electronic map into target map data.

In the embodiment of the application, a map data updating device is provided, and by adopting the device, anchor points can be constructed on an electronic map based on a large amount of vehicle data, and further combined with the anchor points and the large amount of vehicle data to generate and fuse the driving tracks, namely the driving tracks can be optimized by utilizing the fixed anchor points, so that more accurate map data can be obtained.

Alternatively, on the basis of the embodiment corresponding to fig. 17, in another embodiment of the map data updating apparatus 20 provided in the embodiment of the present application,

the obtaining module 201 is specifically configured to receive vehicle data corresponding to a vehicle if the vehicle turns before passing through a target node or after passing through the target node, where the target node is any node included in the target original map data, and the vehicle is one or more vehicles among the N vehicles.

In the embodiment of the application, the map data updating device is provided, and by adopting the device, the position of the vehicle can be determined in real time based on the positioning system, so that whether the vehicle turns in advance or turns in a delayed manner at a place needing to turn can be detected, and whether the vehicle data of the vehicle is uploaded to the server or not is determined. If the vehicle does not turn ahead or turn behind, the vehicle data does not need to be reported to the server, so that resources consumed by data transmission are saved, if the vehicle turns ahead or turns behind, the vehicle data is automatically reported so that the server can update the map data based on the data, when enough vehicles pass through the same intersection, the map data can be updated according to the driving track, and therefore more accurate map data can be obtained.

a generating module 202, specifically configured to obtain an area of interest according to an image to be identified included in the vehicle data;

acquiring characteristic points corresponding to the region of interest;

and determining the position corresponding to the anchor point in the electronic map according to the three-dimensional expansion feature information corresponding to the region of interest. And determining the position corresponding to the anchor point in the electronic map according to the three-dimensional expansion feature information corresponding to the region of interest.

In the embodiment of the application, the map data updating device is provided, and by adopting the device, the positions of anchor points under a world coordinate system can be fitted according to vehicle data reported by a large number of vehicles, so that the situation that the position deviation of the anchor points is large due to the fact that the quantity is small or the positioning data is not accurate enough is avoided, and the accuracy of the positions of the anchor points is improved.

the generating module 202 is specifically configured to obtain an image recognition result through an image recognition model based on an image to be recognized included in the vehicle data;

In the embodiment of the application, a map data updating device is provided, and by adopting the device, the trained image recognition model can be used for rapidly and accurately extracting the main region of interest and the sub region of interest in the image to be recognized, and further recognizing the object type in the region of interest, so that the selection accuracy of the anchor point can be improved, and the positioning reliability can be improved.

the generating module 202 is specifically configured to determine three-dimensional extended feature information corresponding to the region of interest as three-dimensional extended feature information corresponding to the anchor point to be matched;

In the embodiment of the application, a map data updating device is provided, and by adopting the device, after the candidate anchor point is obtained, whether the candidate anchor point is successfully matched with the existing candidate anchor point needs to be judged, if the matching is successful, the candidate anchor point is represented as a real anchor point, and if the matching is failed, the candidate anchor point is used as a new candidate anchor point for subsequent matching. Therefore, candidate anchor points are added in an incremental matching mode, on one hand, more accurate anchor points can be selected, on the other hand, full matching is not needed each time, and resource consumption in the matching process can be reduced.

the generating module 202 is specifically configured to determine a matching degree according to three-dimensional extended feature information corresponding to an anchor point to be matched and three-dimensional extended feature information corresponding to a first candidate anchor point;

In the embodiment of the application, a map data updating device is provided, and by adopting the device, because vehicle data contains data reported by a large number of vehicles, differences of suggested devices may have some data with large errors, the data belong to 'noise' in a data processing process, and anchors to be matched with large errors need to be filtered in order to improve the accuracy of anchor positions, so that the reliability of anchor calculation is improved.

the generating module 202 is specifically configured to generate multidimensional gaussian distribution corresponding to the electronic map according to the vehicle data;

In the embodiment of the application, a map data updating device is provided, and by adopting the device, an anchor point is fixed, then the node position on a fusion driving track is adjusted by adopting a light beam adjustment algorithm, and finally a more accurate fusion driving track is obtained, so that the feasibility and the reliability of the scheme are improved.

the generating module 202 is specifically configured to construct a two-dimensional original map corresponding to the electronic map, where the two-dimensional original map includes (N × M) grids, and N and M are both integers greater than 1;

In the embodiment of the application, a map data updating device is provided, and by adopting the device, the position parameters of different vehicles at different moments are extracted from vehicle data reported by a large number of vehicles, then the optimized local driving tracks are recorded in a two-dimensional original map according to the position parameters of each vehicle at different moments, and finally, corresponding two-dimensional driving probability maps are generated according to the local driving tracks of all vehicles on the two-dimensional original map, so that the effect of fusing the optimized local driving tracks is achieved, and the authenticity and the reliability of the two-dimensional driving probability maps are improved.

the generating module 202 is specifically configured to obtain a length in the primary direction and a length in the secondary direction according to the multidimensional gaussian distribution;

and determining the lane type according to the segmentation threshold.

In the embodiment of the application, a map data updating device is provided, and by adopting the device, the corresponding flat parameters can be generated by combining the specific situation of multi-dimensional Gaussian distribution, and the segmentation threshold value is determined based on the flat parameters, so that a more reasonable segmentation threshold value can be generated, and the purpose of dynamically adjusting the segmentation threshold value is realized.

the generating module 202 is specifically configured to perform binarization processing on the two-dimensional driving probability map by using a segmentation threshold to obtain a binarization driving probability map, where the binarization driving probability map includes a first region, and a probability value of each pixel point in the first region is greater than or equal to the segmentation threshold;

In the embodiment of the application, the map data updating device is provided, and after the segmentation threshold is determined, binarization processing can be performed on the two-dimensional driving probability map based on the segmentation threshold, so that all pixel points larger than or equal to the segmentation threshold are selected, and then a second area is obtained in the two-dimensional driving probability map, so that a merging ratio can be determined, the lane type can be more accurately judged based on the merging ratio, and feasibility and operability of a scheme are improved.

the obtaining module 201 is specifically configured to extract Q updated nodes from the fused vehicle trajectory, where Q is an integer greater than 1, and the Q updated nodes at least include nodes before turning and nodes after turning.

In the embodiment of the application, the device is adopted, binarization processing is performed on the two-dimensional driving probability map by adopting a segmentation threshold value to obtain a binarization driving probability map, under the condition that the lane type is determined to be a single lane type, curve fitting processing is performed on a first area included in the binarization driving probability map, namely, a fused driving track can be generated, the fused driving track is optimized based on a beam adjustment algorithm to obtain the fused driving track, therefore, at least two updated nodes can be extracted from the fused driving track, and the feasibility and the operability of the scheme are improved.

generating a fused driving track according to the polygonal area;

In the embodiment of the application, a map data updating device is provided, and by adopting the device, a segmentation threshold value is adopted to carry out binarization processing on a two-dimensional driving probability map to obtain a binarization driving probability map, under the condition that the lane type is determined to be a multi-lane type, a polygonal area is formed according to a first area included in the binarization driving probability map, then a skeleton in the polygonal area is extracted, so that a fitted and processed fused driving track is obtained, the fused driving track is optimized based on a beam adjustment algorithm to obtain a fused driving track, so that at least two updated nodes can be extracted from the fused driving track, and the feasibility and operability of the scheme are improved.

Fig. 18 is a schematic diagram of a server 300 according to an embodiment of the present disclosure, where the server 300 may have a relatively large difference due to different configurations or performances, and may include one or more Central Processing Units (CPUs) 322 (e.g., one or more processors) and a memory 332, and one or more storage media 330 (e.g., one or more mass storage devices) for storing applications 342 or data 344. Memory 332 and storage media 330 may be, among other things, transient storage or persistent storage. The program stored on the storage medium 330 may include one or more modules (not shown), each of which may include a series of instruction operations for the server. Still further, the central processor 322 may be configured to communicate with the storage medium 330 to execute a series of instruction operations in the storage medium 330 on the server 300.

The Server 300 may also include one or more power supplies 326, one or more wired or wireless network interfaces 350, one or more input-output interfaces 358, and/or one or more operating systems 341, such as a Windows Server^TM，Mac OS X^TM，Unix^TM, Linux^TM，FreeBSD^TMEtc. of。

The steps performed by the server in the above embodiment may be based on the server structure shown in fig. 18.

Embodiments of the present application also provide a computer-readable storage medium and a computer program product, where the computer program is stored in the computer-readable storage medium, and when the computer program runs on a computer, the computer is caused to execute the method described in the foregoing embodiments. When the computer program product runs on a computer, the computer is caused to execute the methods described in the foregoing embodiments.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be substantially implemented or contributed to by the prior art, or all or part of the technical solution may be embodied in a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.

The above embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.

Claims

1. A method for map data update, comprising:

generating an anchor point according to the vehicle data;

generating a fused driving track according to the anchor point and the vehicle data;

and updating target original map data in the electronic map into the target map data.

2. The method of claim 1, wherein the obtaining vehicle data comprises:

and if the vehicle turns before passing through the target node or after passing through the target node, receiving vehicle data corresponding to the vehicle, wherein the target node is any node included in the target original map data, and the vehicle is one or more vehicles in the N vehicles.

3. The method of claim 1, wherein the generating anchor points from the vehicle data comprises:

acquiring an interested area according to the image to be identified included in the vehicle data;

acquiring a characteristic point corresponding to the region of interest;

generating extended characteristic information according to the characteristic points corresponding to the region of interest and the direction parameters and the position parameters included in the vehicle data;

generating three-dimensional extended characteristic information corresponding to the region of interest according to the characteristic points corresponding to the region of interest and the extended characteristic information;

4. The method according to claim 3, wherein the acquiring a region of interest from the image to be identified included in the vehicle data comprises:

acquiring an image recognition result through an image recognition model based on the image to be recognized included in the vehicle data;

if the existence of the indication mark is determined according to the image identification result, acquiring a main region of interest, wherein the main region of interest belongs to the region of interest;

and if the existence of the static reference object is determined according to the image identification result, acquiring a secondary region of interest, wherein the secondary region of interest belongs to the region of interest.

5. The method according to claim 3, wherein the determining, in the electronic map, the position corresponding to the anchor point according to the three-dimensional extended feature information corresponding to the region of interest includes:

determining the three-dimensional expansion characteristic information corresponding to the region of interest as the three-dimensional expansion characteristic information corresponding to the anchor point to be matched;

and if the three-dimensional extended feature information corresponding to the anchor point to be matched fails to be matched with the three-dimensional extended feature information corresponding to the first candidate anchor point, determining the anchor point to be matched as a second candidate anchor point.

6. The method of claim 5, wherein the matching the three-dimensional extended feature information corresponding to the anchor point to be matched with the three-dimensional extended feature information corresponding to the first candidate anchor point comprises:

determining the matching degree according to the three-dimensional expansion feature information corresponding to the anchor point to be matched and the three-dimensional expansion feature information corresponding to the first candidate anchor point;

after the three-dimensional extended feature information corresponding to the anchor point to be matched is matched with the three-dimensional extended feature information corresponding to the first candidate anchor point, the method further includes:

if the matching degree is greater than or equal to a matching degree threshold value, determining that the three-dimensional extended feature information corresponding to the anchor point to be matched is successfully matched with the three-dimensional extended feature information corresponding to the first candidate anchor point;

and if the matching degree is less than or equal to the threshold value of the matching degree, filtering the anchor point to be matched.

7. The method of any one of claims 1 to 6, wherein generating a fused trajectory from the anchor point and the vehicle data comprises:

generating multidimensional Gaussian distribution corresponding to the electronic map according to the vehicle data;

and generating the fused vehicle track according to the lane type and the two-dimensional driving probability map.

8. The method of claim 7, wherein generating the two-dimensional driving probability map corresponding to the electronic map according to the anchor point and the vehicle data comprises:

constructing a two-dimensional original map corresponding to the electronic map, wherein the two-dimensional original map comprises (N multiplied by M) grids, and both N and M are integers greater than 1;

acquiring K local driving tracks according to the position parameters included in the vehicle data, wherein the vehicle data include vehicle data of K vehicles, the vehicle data of each vehicle are used for generating one local driving track, and K is an integer greater than 1;

according to the anchor points, optimizing each local trajectory in the K local trajectories to obtain K optimized local trajectories;

and mapping the K optimized local driving tracks to the two-dimensional original graph to obtain the two-dimensional driving probability graph, wherein the probability value corresponding to the grid in the two-dimensional driving probability graph is in positive correlation with the passing times of the K optimized local driving tracks.

9. The method of claim 7, wherein determining the lane type from the multi-dimensional Gaussian distribution comprises:

acquiring the length of a main direction and the length of a secondary direction according to the multi-dimensional Gaussian distribution;

generating a flat parameter according to the length of the main direction and the length of the secondary direction;

determining a segmentation threshold according to the flattening parameters;

and determining the lane type according to the segmentation threshold.

10. The method of claim 9, wherein said determining the lane type according to the segmentation threshold comprises:

performing binarization processing on the two-dimensional driving probability map by using the segmentation threshold value to obtain a binarization driving probability map, wherein the binarization driving probability map comprises a first region, and the probability value of each pixel point in the first region is greater than or equal to the segmentation threshold value;

if the intersection ratio is smaller than or equal to an intersection ratio threshold value, determining that the lane type is the single lane type;

and if the intersection ratio is larger than the intersection ratio threshold value, determining that the lane type is the multi-lane type.

11. The method of claim 7, wherein generating a fused trajectory from the lane type and the two-dimensional driving probability map comprises:

if the lane type is the single lane type, performing curve fitting processing on the first area included in the binarization driving probability map to generate the fused driving track;

the acquiring of the target map data according to the fused driving track comprises the following steps:

and extracting Q updated nodes from the fused driving track, wherein Q is an integer larger than 1, and the Q updated nodes at least comprise nodes before turning and nodes after turning.

12. The method of claim 7, wherein generating a fused trajectory from the lane type and the two-dimensional driving probability map comprises:

if the lane type is the multi-lane type, constructing a polygonal area according to the first area included in the binarization driving probability map;

generating the fused vehicle path according to the polygonal area;

13. A map data updating apparatus, comprising:

the generating module is further used for generating a fused driving track according to the anchor point and the vehicle data;

and the updating module is used for updating the target original map data in the electronic map into the target map data.

14. A server, comprising: a memory, a processor, and a bus system;

wherein the memory is used for storing programs;

the processor for executing the program in the memory, the processor for performing the method of any one of claims 1 to 12 according to instructions in program code;

15. A computer-readable storage medium comprising instructions that, when executed on a computer, cause the computer to perform the method of any of claims 1 to 12.