Training method and device of path planning model, equipment and storage medium

Abstract

The invention discloses a training method and device of a path planning model, equipment and a storage medium, and relates to the technical field of computers. The method comprises the following steps: acquiring sample data of a sample aerial view feature map including sample vehicle driving data and sensing information corresponding to the sample vehicle driving data; encoding the sample aerial view feature map by using a first encoding network of a to-be-trained path planning model to obtain aerial view feature map encoding information; encoding the sample vehicle driving data by using a second encoding network of the path planning model to obtain vehicle driving feature encoding information; splicing the aerial view feature map coding information and the vehicle driving feature coding information to obtain fused feature coding information; decoding the fusion feature coding information by using a decoding network of the path planning model to obtain a path planning result; and based on the sample data and the path planning result, performing training processing on the path planning model to obtain a trained path planning model.

Description

Technical Field

[0001] The present application relates to the field of computer technology, specifically to technical fields such as intelligent transportation and autonomous driving, and in particular to a training method, device, equipment and storage medium for a path planning model. Background Art

[0002] Generally, the path planning system is one of the core systems of autonomous driving technology. Its main task is to generate a better driving route for the autonomous driving vehicle based on perception information, prediction information, and high-precision map information.

[0003] At present, the path planning system in related technologies is usually divided into three sub-modules: behavior decision, path planning and motion planning. Among them, the behavior decision module is responsible for determining the high-level operations that the vehicle should perform in a specific situation. The task of the path planning module is to generate a global path from the current position of the vehicle to the target position, usually taking into account the map information and the output of the behavior decision module. The motion planning module is responsible for converting the global path generated by the path planning into a specific executable trajectory, including dynamic parameters such as acceleration, speed and steering angle.

[0004] However, there are still some problems with the path planning system in related technologies: on the one hand, the entire path planning system is divided into several subsystems and multiple functional modules, and each module is affected by computing power, resulting in a low performance ceiling. On the other hand, the path planning system is overly dependent on high-precision maps, making data calculation and processing more difficult. Summary of the invention

[0005] The present application provides a training method, device, equipment and storage medium for a path planning model, which solves the problems of poor performance of the path planning system. The technical solution is as follows:

[0006] In a first aspect, a method for training a path planning model is provided, wherein the path planning model to be trained includes a first encoding network, a second encoding network, and a decoding network, and the method includes:

[0007] Acquire sample data, the sample data including sample vehicle driving data and a sample bird's-eye view feature map of perception information corresponding to the sample vehicle driving data;

[0008] Using the first coding network, encoding the sample bird's-eye view feature image to obtain bird's-eye view feature image encoding information;

[0009] Using the second coding network, encoding the sample vehicle driving data to obtain vehicle driving feature coding information;

[0010] Performing feature splicing processing on the bird's-eye view feature image coding information and the vehicle driving feature coding information to obtain fused feature coding information;

[0011] Using the decoding network, decoding the fused feature encoding information to obtain a path planning result;

[0012] Based on the sample data and the path planning result, the path planning model to be trained is updated and trained to obtain a trained path planning model.

[0013] In a possible implementation, the first encoding network includes a spatial feature encoder and a first position encoder, and using the first encoding network to encode the sample bird's-eye view feature map to obtain bird's-eye view feature map encoding information includes:

[0014] Using the spatial feature encoder, performing spatial feature encoding processing on the sample bird's-eye view feature image to obtain a result of the spatial feature encoding processing of the sample bird's-eye view feature image;

[0015] Using the first position encoder, performing position encoding processing on the sample bird's-eye view feature map to obtain a result of the position encoding processing of the sample bird's-eye view feature map;

[0016] The result of the spatial feature encoding process and the result of the position encoding process are added together to obtain the bird's-eye view feature map encoding information.

[0017] In a possible implementation, the second coding network includes a driving state feature encoder and a second position encoder, and the encoding process of the sample vehicle driving data using the second coding network to obtain vehicle driving feature coding information includes:

[0018] Using the driving state feature encoder, performing driving state feature encoding processing on the sample vehicle driving data to obtain a result of the driving state feature encoding processing of the sample vehicle driving data;

[0019] Using the second position encoder, performing position encoding processing on the sample vehicle driving data to obtain a result of the position encoding processing of the sample vehicle driving data;

[0020] The result of the driving state feature coding process and the result of the position coding process are added together to obtain the vehicle driving feature coding information.

[0021] In a possible implementation, the decoding network includes a decoder, a classification fully connected layer, and a regression fully connected layer, and the decoding network is used to decode the fusion feature encoding information to obtain a path planning result, including:

[0022] Using the decoder, decoding the fused feature encoding information to obtain a decoding result;

[0023] Inputting the result of the decoding process into the classification fully connected layer to obtain the driving direction category information of the path planning;

[0024] Inputting the result of the decoding process into the regression fully connected layer to obtain trajectory information of the path planning;

[0025] The path planning result is obtained based on the driving direction category information and the trajectory information.

[0026] In a second aspect, a path planning method is provided, the method comprising:

[0027] Acquire vehicle driving data to be processed and a bird's-eye view feature map of perception information corresponding to the vehicle driving data;

[0028] Inputting the vehicle driving data and the bird's-eye view feature map into a pre-trained path planning model to obtain a result output by the path planning model;

[0029] Obtaining a path planning result based on a result output by the path planning model;

[0030] The path planning model is pre-trained using the method according to the aspect and any possible implementation method described in the claim, and the path planning model includes a first encoding network, a second encoding network and a decoding network.

[0031] In a third aspect, a training device for a path planning model is provided, wherein the path planning model to be trained includes a first encoding network, a second encoding network and a decoding network, and the device includes:

[0032] A first acquisition unit is used to acquire sample data, wherein the sample data includes sample vehicle driving data and a sample bird's-eye view feature map of perception information corresponding to the sample vehicle driving data;

[0033] A first encoding unit, configured to use the first encoding network to encode the sample bird's-eye view feature image to obtain bird's-eye view feature image encoding information;

[0034] A second encoding unit is used to encode the sample vehicle driving data using the second encoding network to obtain vehicle driving feature encoding information;

[0035] A first splicing unit is used to perform feature splicing processing on the bird's-eye view feature image coding information and the vehicle driving feature coding information to obtain fused feature coding information;

[0036] A first decoding unit, configured to use the decoding network to decode the fused feature encoding information to obtain a path planning result;

[0037] The first training unit is used to update the path planning model to be trained based on the sample data and the path planning result to obtain a trained path planning model.

[0038] In a fourth aspect, a path planning device is provided, the device comprising:

[0039] An acquisition unit, used to acquire vehicle driving data to be processed and a bird's-eye view feature map of perception information corresponding to the vehicle driving data;

[0040] A planning unit, used for inputting the vehicle driving data and the bird's-eye view feature map into a pre-trained path planning model to obtain a result output by the path planning model;

[0041] An obtaining unit, configured to obtain a path planning result based on a result output by the path planning model;

[0042] Wherein, the path planning model is pre-trained using the device described in the third aspect above, and the path planning model includes a first encoding network, a second encoding network and a decoding network.

[0043] In a fifth aspect, a computer-readable storage medium is provided, wherein the storage medium stores at least one instruction, and the at least one instruction is loaded and executed by a processor to implement the method of the above-mentioned aspects and any possible implementation manner.

[0044] In a sixth aspect, an electronic device is provided, including:

[0045] at least one processor; and

[0046] a memory communicatively connected to the at least one processor; wherein,

[0047] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any possible implementation manner and the aspects described above.

[0048] In a seventh aspect, an autonomous driving vehicle is provided, comprising the electronic device as described above.

[0049] The beneficial effects of the technical solution provided by this application include at least:

[0050] It can be seen from the above technical scheme that, on the one hand, the embodiment of the present application can obtain sample data, wherein the sample data includes sample vehicle driving data and a sample bird's-eye view feature map of the perception information corresponding to the sample vehicle driving data, and use the first coding network of the path planning model to be trained to encode the sample bird's-eye view feature map to obtain the bird's-eye view feature map encoding information, use the second coding network of the path planning model to be trained to encode the sample vehicle driving data to obtain the vehicle driving feature encoding information, perform feature splicing processing on the bird's-eye view feature map encoding information and the vehicle driving feature encoding information to obtain fused feature encoding information, and use the path planning model to be trained to obtain the fused feature encoding information. The decoding network of the planning model decodes the fused feature coding information to obtain a path planning result, and based on the sample data and the path planning result, updates the training process of the path planning model to be trained to obtain a trained path planning model. Since the path planning model including the first encoding network, the second encoding network and the decoding network is trained based on the sample vehicle driving data and the sample bird's-eye view feature map of the perception information corresponding to the sample vehicle driving data, a more accurate and effective end-to-end path planning model can be obtained, which ensures the performance and effectiveness of the path planning model, thereby ensuring the reliability and performance of the path planning of the autonomous driving vehicle.

[0051] It can be seen from the above technical solution that, on the other hand, the embodiment of the present application can obtain the vehicle driving data to be processed and the bird's-eye feature map of the perception information corresponding to the vehicle driving data, and input the vehicle driving data and the bird's-eye feature map into the path planning model to obtain the path planning result. The path planning model includes a first encoding network, a second encoding network and a decoding network. Since the pre-trained path planning model has better end-to-end path planning performance, the end-to-end path planning model is used to perform path planning on the acquired vehicle driving data and the bird's-eye feature map of the perception information corresponding to the vehicle driving data. While ensuring the accuracy of the path planning result, the corresponding path planning result can be directly output based on the end-to-end planning model, which reduces the computing power overhead of path planning, optimizes the planning performance and efficiency, and thus improves the reliability of path planning of autonomous driving vehicles.

[0052] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0054] Figure 1 It is a flowchart of a training method for a path planning model provided by an embodiment of the present application;

[0055] Figure 2 is a flowchart of a path planning method provided by another embodiment of the present application;

[0056] Figure 3 is a flowchart of a path planning method provided by another embodiment of the present application;

[0057] Figure 4A is a schematic diagram of drivable area information in a bird's-eye view feature map in a path planning method provided by another embodiment of the present application;

[0058] Figure 4B is a schematic diagram of occupancy information in a bird's-eye view feature map in a path planning method provided by another embodiment of the present application;

[0059] Figure 5 It is a schematic diagram of the architecture of an application scenario of a path planning method provided by another embodiment of the present application;

[0060] Figure 6 is a structural block diagram of a training device for a path planning model provided by another embodiment of the present application;

[0061] Figure 7 is a structural block diagram of a path planning device provided by another embodiment of the present application;

[0062] Figure 8 It is a block diagram of an electronic device used to implement the path planning method and model training method of the embodiments of the present disclosure. DETAILED DESCRIPTION

[0063] The following is a description of exemplary embodiments of the present application in conjunction with the accompanying drawings, including various details of the embodiments of the present application to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0064] Obviously, the described embodiments are only part of the embodiments of the present application, but not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0065] It should be noted that the terminal devices involved in the embodiments of the present application may include but are not limited to mobile phones, personal digital assistants (PDAs), wireless handheld devices, tablet computers and other smart devices; display devices may include but are not limited to personal computers, televisions and other devices with display functions.

[0066] In addition, the term "and / or" in this article is only a description of the association relationship between the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0067] At present, the path planning module is usually divided into three sub-modules: behavior decision, path planning and motion planning. Among them, the behavior decision module is responsible for determining the high-level operations that the vehicle should perform in a specific situation. It decides whether the vehicle should accelerate, decelerate, change lanes, stop or turn based on the environmental data perceived by the sensors, such as the position, speed, pedestrians, etc. of other vehicles, as well as traffic rules. The task of the path planning module is to generate a global path from the current position of the vehicle to the target position, usually taking into account map information, such as road networks, lane structures, etc. and the output of the behavior decision module. Path planning defines a series of discrete path points or curves to ensure that the vehicle follows the predetermined route. The motion planning module is responsible for converting the global path generated by path planning into a specific executable trajectory, including dynamic parameters such as acceleration, speed and steering angle. This part of the planning needs to meet the vehicle's dynamic constraints and the safety requirements of the environment. In addition to the three sub-modules, the current planning system generally also includes some sub-functions, such as initialization, scene management, task management, horizontal and vertical separation, trajectory post-processing, etc.

[0068] However, there are some major problems with the path planning module in the related technology: on the one hand, the path planning system is complex and has a low performance ceiling: the entire path planning system is divided into several subsystems and functional modules, and each module is affected by computing power, resulting in a low performance ceiling. The complex logic of information transmission between subsystems leads to a large increase in system development, maintenance, and optimization costs. On the other hand, the path planning system is overly dependent on high-precision maps: the creation, update, and maintenance of high-precision maps are difficult and costly; the system's adaptability in areas without high-precision maps deteriorates; the map lacks real-time adaptability, which affects efficiency and safety; high-precision map storage and transmission requires a lot of space, resulting in high requirements for system hardware and communication, making calculation and processing more difficult. Moreover, in general, the system needs to extract a large amount of information from the map, including waypoint location, waypoint direction, polygon center, lane line category, road centerline, traffic light status, speed limit, road topology, etc., and combine them together through complex coding methods. This greatly increases the difficulty of module development, slows down the development cycle of the autonomous driving system, and thus affects the implementation of high-level autonomous driving systems.

[0069] Therefore, there is an urgent need for a path planning model training method that can improve the performance of path planning and thus ensure the reliability of path planning for autonomous driving vehicles.

[0070] Please refer to Figure 1 , which shows a flow chart of a training method for a path planning model provided by an embodiment of the present application. The path planning model to be trained may include a first encoding network, a second encoding network and a decoding network. The training method for the path planning model may specifically include:

[0071] Step 101: Acquire sample data, where the sample data includes sample vehicle driving data and a sample bird's-eye view feature map of perception information corresponding to the sample vehicle driving data.

[0072] Step 102: Utilize the first coding network to perform coding processing on the sample bird's-eye view feature map to obtain bird's-eye view feature map coding information.

[0073] Step 103: Utilize the second coding network to encode the sample vehicle driving data to obtain vehicle driving characteristic coding information.

[0074] Step 104: perform feature splicing processing on the bird's-eye view feature image coding information and the vehicle driving feature coding information to obtain fused feature coding information.

[0075] Step 105: Utilize the decoding network to decode the fused feature encoding information to obtain a path planning result.

[0076] Step 106: Based on the sample data and the path planning result, update the path planning model to be trained to obtain a trained path planning model.

[0077] It should be noted that the path planning model to be trained can be a model based on a neural network.

[0078] It should be noted that the sample vehicle driving data may include but is not limited to the position, speed, acceleration, direction, etc. of the autonomous driving vehicle.

[0079] It should be noted that the sample bird's-eye view feature map can be a sample bird's-eye view feature map determined based on perception information. The sample bird's-eye view feature map can be rasterized perception information generated based on a bird's-eye view (BEV) spatial representation. The sample bird's-eye view feature map can be directly obtained from an upstream perception module. The sample bird's-eye view feature map may include but is not limited to drivable area information, route information, and occupancy information.

[0080] It should be noted that part or all of the execution subject of steps 101 to 106 may be an application located in the local terminal, or may also be a functional unit such as a plug-in or software development kit (SDK) set in the application located in the local terminal, or may also be a processing engine located in a network-side server, or may also be a distributed system located on the network side, for example, a processing engine or distributed system in a model training platform on the network side, etc. This embodiment does not specifically limit this.

[0081] It is understandable that the application may be a local program (nativeApp) installed on the local terminal, or may be a webpage program (webApp) of a browser on the local terminal, which is not limited in this embodiment.

[0082] In this way, the path planning model including the first encoding network, the second encoding network and the decoding network can be trained and processed by using a sample bird's-eye view feature map based on the perception information corresponding to the sample vehicle driving data, so that a more accurate and effective end-to-end path planning model can be obtained, thereby ensuring the performance and effectiveness of the path planning model, thereby ensuring the reliability and performance of the path planning of the autonomous driving vehicle.

[0083] Moreover, the sample data used to train the path planning model does not involve high-precision map data, which can further reduce the amount of data to be processed for model training. While ensuring the performance and effectiveness of the path planning model, it can reduce the processing computing overhead of the overall model training, thereby improving the reliability and efficiency of path planning.

[0084] Optionally, in a possible implementation of this embodiment, the first coding network may include a spatial feature encoder and a first position encoder. In step 102, first, the spatial feature encoder may be used to perform spatial feature coding processing on the sample bird's-eye view feature map to obtain the result of the spatial feature coding processing of the sample bird's-eye view feature map. Secondly, the first position encoder may be used to perform position coding processing on the sample bird's-eye view feature map to obtain the result of the position coding processing of the sample bird's-eye view feature map. Thirdly, the result of the spatial feature coding processing and the result of the position coding processing are added to obtain the coding information of the bird's-eye view feature map.

[0085] In this implementation, the spatial feature encoder may be an encoder based on a convolutional neural network (CNN). The spatial feature encoder may be used to process the spatial feature information in the bird's-eye view feature map of the sample.

[0086] In a specific implementation process of this implementation method, the sample bird's-eye view feature map can be input into a CNN-based encoder, and the result of the spatial feature encoding processing of the sample bird's-eye view feature map can be output.

[0087] Here, the spatial feature information in the sample bird's-eye view feature map may include edge information and shape information of the relevant area.

[0088] In this implementation, the first position encoder may be an encoder based on a multi-layer perceptron. The first position encoder may be used to process position information in the sample bird's-eye view feature map.

[0089] In another specific implementation process of this implementation method, the sample bird's-eye view feature map can be input into a position encoder based on a multi-layer perceptron, and the result of the position encoding processing of the sample bird's-eye view feature map can be output.

[0090] Here, the location information in the sample bird's-eye view feature map may include coordinate information, etc.

[0091] In this way, the input sample bird's-eye view feature map can be encoded by using the spatial feature encoder and the first position encoder, respectively, to obtain the corresponding spatial feature encoding processing results and position encoding processing results, which can further improve the accuracy of the encoding processing results of the sample bird's-eye view feature map.

[0092] Optionally, in a possible implementation of this embodiment, the second coding network may include a driving state feature encoder and a second position encoder. In step 103, first, the driving state feature encoder may be used to perform driving state feature encoding processing on the sample vehicle driving data to obtain the result of the driving state feature encoding processing of the sample vehicle driving data. Secondly, the second position encoder may be used to perform position encoding processing on the sample vehicle driving data to obtain the result of the position encoding processing of the sample vehicle driving data. Thirdly, the result of the driving state feature encoding processing and the result of the position encoding processing may be added to obtain the vehicle driving feature encoding information.

[0093] In this implementation, the driving state feature encoder may be an encoder based on a multi-layer perceptron. The driving state feature encoder may be used to process driving state information in the sample vehicle driving data. The driving state information of the autonomous driving vehicle may include speed, acceleration, direction, etc.

[0094] In this implementation, the second position encoder may also be an encoder based on a multi-layer perceptron. The second position encoder may be used to process position information in vehicle driving data of the autonomous driving vehicle. The second position encoder information of the autonomous driving vehicle may include coordinate information, etc.

[0095] In this way, the input sample vehicle driving data can be encoded by using the driving state feature encoder and the second position encoder, respectively, to obtain the corresponding driving state feature encoding processing results and position encoding processing results, which can further improve the accuracy of the obtained vehicle driving feature coding information.

[0096] It should be noted that the specific implementation process provided in this implementation can be combined with the various specific implementation processes provided in the aforementioned implementation to implement the training method of the path planning model of this embodiment. For a detailed description, please refer to the relevant content in the aforementioned implementation, which will not be repeated here.

[0097] Optionally, in a possible implementation of this embodiment, the decoding network may include a decoder, a classification fully connected layer, and a regression fully connected layer. In step 105, first, the decoder may be used to decode the fused feature encoding information to obtain a decoding result. Secondly, the decoding result may be input into the classification fully connected layer to obtain the driving direction category information of the path planning. Thirdly, the decoding result may be input into the regression fully connected layer to obtain the trajectory information of the path planning. Thirdly, the path planning result may be obtained based on the driving direction category information and the trajectory information.

[0098] In this implementation, the decoder may be a decoder based on a multi-layer transform structure. Preferably, the decoder may be a decoder based on a 4-layer transform structure.

[0099] In this implementation, the driving direction category information may include straight ahead, left turn, right turn, etc. The trajectory information may include the three-dimensional coordinate information and speed of the planned trajectory points.

[0100] In a specific implementation process of this implementation method, the driving direction category information and the trajectory information corresponding to the driving direction category information can be used as a path planning result.

[0101] For example, for one planning cycle, the path planning result of the planning cycle may include going straight and a series of trajectory points corresponding to going straight. For another planning cycle, the path planning result of the planning cycle may include turning left and a series of trajectory points corresponding to turning left.

[0102] In this way, the fused feature encoding information can be decoded by using a decoder, a classification fully connected layer, and a regression fully connected layer to obtain the corresponding path planning results including the driving direction category information and trajectory information, which can further improve the accuracy of the model output results.

[0103] It should be noted that the specific implementation process provided in this implementation can be combined with the various specific implementation processes provided in the aforementioned implementation to implement the training method of the path planning model of this embodiment. For a detailed description, please refer to the relevant content in the aforementioned implementation, which will not be repeated here.

[0104] Optionally, in a possible implementation of this embodiment, the path planning model to be trained further includes a splicing network. In step 104, the splicing network can be used to perform feature splicing processing on the bird's-eye view feature map encoding information and the vehicle driving feature encoding information to obtain a result of the feature splicing processing, and then based on the result of the feature splicing processing, fusion feature encoding information can be obtained.

[0105] In this implementation, the concatenated network may be a concatenated layer (Concat) network.

[0106] It should be noted that the specific implementation process provided in this implementation can be combined with the various specific implementation processes provided in the aforementioned implementation to implement the training method of the path planning model of this embodiment. For a detailed description, please refer to the relevant content in the aforementioned implementation, which will not be repeated here.

[0107] Figure 2is a flow chart of a path planning method provided by another embodiment of the present application, such as Figure 2 shown.

[0108] Step 201: Acquire vehicle driving data to be processed and a bird's-eye view feature map of perception information corresponding to the vehicle driving data.

[0109] Step 202: input the vehicle driving data and the bird's-eye view feature map into a pre-trained path planning model to obtain a result output by the path planning model.

[0110] Step 203: Obtain a path planning result based on the result output by the path planning model; wherein the path planning model includes a first encoding network, a second encoding network and a decoding network.

[0111] It should be noted that the path planning model is pre-trained using the training method of the path planning model in the above-mentioned embodiment. The path planning model may be a model based on a neural network.

[0112] It should be noted that vehicle driving data may include but is not limited to the position, speed, acceleration, direction, etc. of the autonomous driving vehicle.

[0113] It should be noted that the bird's-eye view feature map can be determined based on perception information. The bird's-eye view feature map can be rasterized perception information generated based on the bird's-eye view (BEV) spatial representation. The bird's-eye view feature map can be directly obtained from the upstream perception module. The bird's-eye view feature map may include but is not limited to drivable area information, route information, and occupancy information.

[0114] It should be noted that part or all of the execution subject of steps 201 to 203 may be an application located in the local terminal, or may also be a functional unit such as a plug-in or software development kit (SDK) set in the application located in the local terminal, or may also be a processing engine located in a network-side server, or may also be a distributed system located on the network side, for example, a processing engine or distributed system in a path planning platform on the network side, etc. This embodiment does not specifically limit this.

[0115] It is understandable that the application may be a local program (nativeApp) installed on the local terminal, or may be a webpage program (webApp) of a browser on the local terminal, which is not limited in this embodiment.

[0116] In this way, the vehicle driving data to be processed and the bird's-eye view feature map of the perception information corresponding to the vehicle driving data can be obtained, and the vehicle driving data and the bird's-eye view feature map can be input into the path planning model to obtain the path planning result. The path planning model includes a first encoding network, a second encoding network and a decoding network. Since the pre-trained path planning model has better end-to-end path planning performance, the end-to-end path planning model is used to perform path planning on the acquired vehicle driving data and the bird's-eye view feature map of the perception information corresponding to the vehicle driving data. While ensuring the accuracy of the path planning result, the corresponding path planning result can be directly output based on the end-to-end planning model, which reduces computing power overhead, optimizes planning performance and efficiency, and thus improves the reliability of path planning for autonomous driving vehicles.

[0117] Optionally, in a possible implementation of this embodiment, in step 202, first, the sample bird's-eye view feature map is encoded using the first encoding network to obtain bird's-eye view feature map encoding information. Secondly, the sample vehicle driving data is encoded using the second encoding network to obtain vehicle driving feature encoding information. Thirdly, the bird's-eye view feature map encoding information and the vehicle driving feature encoding information are feature concatenated to obtain fused feature encoding information. Thirdly, the fused feature encoding information is decoded using the decoding network to obtain a path planning result.

[0118] In this implementation, the first encoding network may include a spatial feature encoder and a first position encoder.

[0119] In a specific implementation process of this implementation, first, the spatial feature encoder can be used to perform spatial feature encoding processing on the sample bird's-eye view feature map to obtain the result of the spatial feature encoding processing of the sample bird's-eye view feature map. Secondly, the first position encoder can be used to perform position encoding processing on the sample bird's-eye view feature map to obtain the result of the position encoding processing of the sample bird's-eye view feature map. Thirdly, the result of the spatial feature encoding processing and the result of the position encoding processing are added to obtain the encoding information of the bird's-eye view feature map.

[0120] In this implementation, the second encoding network may include a driving state feature encoder and a second position encoder.

[0121] In a specific implementation process of this implementation mode, first, the driving state feature encoder can be used to perform driving state feature encoding processing on the sample vehicle driving data to obtain the driving state feature encoding processing result of the sample vehicle driving data. Secondly, the second position encoder can be used to perform position encoding processing on the sample vehicle driving data to obtain the position encoding processing result of the sample vehicle driving data. Thirdly, the driving state feature encoding processing result and the position encoding processing result can be added to obtain the vehicle driving feature encoding information.

[0122] In this implementation, the decoding network may include a decoder, a classification fully connected layer, and a regression fully connected layer.

[0123] In a specific implementation process of this implementation method, first, the decoder can be used to decode the fused feature encoding information to obtain a decoding result. Secondly, the decoding result can be input into the classification fully connected layer to obtain the driving direction category information of the path planning. Thirdly, the decoding result can be input into the regression fully connected layer to obtain the trajectory information of the path planning. Thirdly, the path planning result can be obtained based on the driving direction category information and the trajectory information.

[0124] In this implementation, the path planning model to be trained also includes a splicing network.

[0125] In a specific implementation process of this implementation method, the stitching network can be used to perform feature stitching processing on the bird's-eye view feature map coding information and the vehicle driving feature coding information to obtain a result of the feature stitching processing, and then based on the result of the feature stitching processing, fused feature coding information can be obtained.

[0126] It should be noted that the various specific implementation processes provided in this implementation can be combined with each other to implement the path planning method of this embodiment. For detailed description, please refer to the relevant content in the above implementation, which will not be repeated here.

[0127] In order to better understand the method of the embodiment of the present application, the method of the embodiment of the present application is described below with reference to the accompanying drawings and specific application scenarios.

[0128] Figure 3 is a flow chart of a path planning method provided by another embodiment of the present application, such as Figure 3 In this embodiment, the path planning method may include:

[0129] Step 301: Obtain sample vehicle driving data and a sample bird's-eye view feature map of perception information corresponding to the sample vehicle driving data.

[0130] In this embodiment, the sample vehicle driving data may include but is not limited to the position, speed, acceleration, direction, etc. of the autonomous driving vehicle.

[0131] In this embodiment, the sample bird's-eye view feature map can be a sample bird's-eye view feature map determined based on perception information. The sample bird's-eye view feature map can be directly obtained from the upstream perception module. The sample bird's-eye view feature map can be rasterized perception information generated based on a bird's-eye view (BEV) spatial representation. The sample bird's-eye view feature map can include but is not limited to drivable area information, route information, and occupancy information. The route information can be vehicle navigation route information determined based on navigation information.

[0132] In this embodiment, the sample bird's-eye view feature map may be a sample bird's-eye view feature map of the perception information at the time corresponding to the sample vehicle driving data, or a sample bird's-eye view feature map of the perception information of the area corresponding to the sample vehicle driving data.

[0133] Step 302: Encode the sample bird's-eye view feature map using the first encoding network of the path planning model to be trained to obtain bird's-eye view feature map encoding information.

[0134] Step 303: Utilize the second encoding network of the path planning model to be trained to encode the sample vehicle driving data to obtain vehicle driving feature encoding information.

[0135] Step 304: perform feature splicing processing on the bird's-eye view feature image coding information and the vehicle driving feature coding information to obtain fused feature coding information.

[0136] Step 305: Use the decoding network of the path planning model to be trained to decode the fused feature encoding information to obtain a path planning result.

[0137] Step 306: Based on the sample vehicle driving data, the sample bird's-eye view feature map of the perception information corresponding to the sample vehicle driving data, and the path planning result, the path planning model to be trained is updated and trained to obtain a trained path planning model.

[0138] Step 307: Acquire the vehicle driving data to be processed and the bird's-eye view feature map of the perception information corresponding to the vehicle driving data.

[0139] In this embodiment, the bird's-eye view feature map may include drivable area information (Drivable Area), route information (Route), and occupancy information (Occupancy). Figure 4A Schematic diagram of the drivable area information in the bird's-eye view feature map in the path planning method provided by another embodiment of the present application. Figure 4AAs shown, the blank area can represent the drivable area information. Figure 4B FIG. 1 is a schematic diagram of occupancy information in a bird's-eye view feature map in a path planning method provided by another embodiment of the present application. Figure 4B As shown, the blank area can represent occupancy information. All information in the bird's-eye view feature map can be generated in real time based on the collected perception information. Route information can be obtained from navigation information.

[0140] In this embodiment, the vehicle driving data, namely, ego information, may include the vehicle's position, speed, acceleration, direction, etc. in the current frame and historical frames.

[0141] Step 308: Input the vehicle driving data and the bird's-eye view feature map into a pre-trained path planning model to obtain a result output by the path planning model, where the path planning model includes a first encoding network, a second encoding network, and a decoding network.

[0142] Step 309: Obtain a path planning result based on the result output by the path planning model.

[0143] In this embodiment, Figure 5 FIG. 1 is a schematic diagram of an application scenario of a path planning method provided by another embodiment of the present application. Figure 5 As shown, in this application scenario, the path planning model may include a spatial feature encoder, a first position encoder, a driving state feature encoder, a second position encoder, a splicing network, a decoding network, a classification fully connected layer, and a regression fully connected layer. After the vehicle driving data and the bird's-eye view feature map are input into the path planning model, the path planning result is output. The path planning result includes the driving direction category information of the path planning and the trajectory information of the path planning. The bird's-eye view feature map may include drivable area information, route information, and occupancy information.

[0144] Exemplarily, the path planning model can be based on an end-to-end path planning neural network. First, at the input end, high-precision map information may not be used. Vehicle driving data and rasterized perception information generated based on BEV spatial representation, i.e., a bird's-eye view feature map, may be used directly. The bird's-eye view feature map may include drivable area information, route information, and occupancy information. All information may be generated in real time. Route information may be obtained from navigation information. Vehicle driving data, i.e., vehicle information, may include the vehicle's current and historical position, speed, acceleration, direction, etc.

[0145] Secondly, the bird's-eye view feature map and vehicle driving data are respectively passed through the encoder (Encoder), and after being added to the embedded features (Embedding, Emb) of the position encoder, the two encoded features are concatenated (Concat) and input into the four-layer Transformer decoder for decoding. Finally, the information output by the decoder is passed through the classification head (CLS) and the regression head (Traj Head) to decode the category of the path planning trajectory and the corresponding trajectory 3D information, and obtain the final path planning result, as shown in Figure 5 shown.

[0146] In this way, by building an end-to-end high-precision map-free autonomous driving path planning model and using millions of data for model training, the entire path planning model system can ultimately achieve excellent performance. The following table shows the performance of various planning models on the test data set. The path planning model of the method of this embodiment is mostly optimal in various closed-loop indicators, such as collision, drivable area, comfort, driving progress, speed, etc.

[0147] Table 1 Comparison of closed-loop indicators of different planning models

[0148] Model Fraction collision Collision time Available area Comfort Driving Progress speed IDM 79.31 90.92 83.49 94.04 94.4 86.16 97.33 PDM-Open 50.24 74.54 69.08 87.89 99.54 69.86 97.72 GC-PGP 61.09 85.87 80.18 89.72 90 60.32 99.72 RasterModel 66.92 86.97 81.46 85.04 81.46 80.6 98.03 UrbanDriver 67.72 85.6 80.28 90.83 100 80.83 91.58 PlanTF 85.3 94.13 90.73 96.79 93.67 89.83 97.78 Path planning model 85.81 95.08 90.42 97.32 97.41 89.35 97.94

[0149] Among them, IDM (Intelligent Driver Model) is an intelligent transportation model. PDM-Open is a prediction drive model (open loop). GC-PGP is a grey correlation (GC)-principal component projection (PCP) coupling model based on nonlinear transformation. Raster Model is a raster data model. Urban Drive is an urban driving model. Plan TF is a simulation-based planner.

[0150] In addition, other planning model solutions all require the use of high-precision maps and traditional modular designs to varying degrees. However, the construction of the path planning model in this embodiment does not involve high-precision maps, so the path planning model of this application has more performance advantages.

[0151] In this embodiment, a new end-to-end high-precision map-free autonomous driving path planning model is proposed here, which replaces the complex traditional path planning system with an entire neural network-based path planning model, removes the dependence on high-precision maps at the input end, and integrates the input of perception and prediction. The entire end-to-end path planning model has a simple structure, a high performance ceiling, simple research and development and maintenance, and fast iteration, which significantly improves the speed of mass production of autonomous driving. In the commercial vehicle industry, it can greatly reduce the cost of autonomous driving productization, laying the foundation for the large-scale promotion of high-end autonomous driving products.

[0152] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0153] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0154] Figure 6 A structural block diagram of a training device for a path planning model provided by an embodiment of the present application is shown. Figure 6 The path planning model training device 600 of this embodiment may include a first acquisition unit 601 , a first encoding unit 602 , a second encoding unit 603 , a first concatenation unit 604 , a first decoding unit 605 and a first training unit 606 . Among them, the path planning model to be trained includes a first encoding network, a second encoding network and a decoding network, a first acquisition unit 601, which is used to obtain sample data, wherein the sample data includes sample vehicle driving data and a sample bird's-eye view feature map of the perception information corresponding to the sample vehicle driving data; a first encoding unit 602, which is used to encode the sample bird's-eye view feature map using the first encoding network to obtain bird's-eye view feature map encoding information; a second encoding unit 603, which is used to encode the sample vehicle driving data using the second encoding network to obtain vehicle driving feature encoding information; a first splicing unit 604, which is used to perform feature splicing processing on the bird's-eye view feature map encoding information and the vehicle driving feature encoding information to obtain fused feature encoding information; a first decoding unit 605, which is used to decode the fused feature encoding information using the decoding network to obtain a path planning result; a first training unit 606, which is used to update the path planning model to be trained based on the sample data and the path planning result to obtain a trained path planning model.

[0155] It should be noted that part or all of the training device of the path planning model of this embodiment may be an application located in the local terminal, or may also be a functional unit such as a plug-in or software development kit (SDK) set in the application located in the local terminal, or may also be a processing engine located in a network-side server, or may also be a distributed system located on the network side, for example, a processing engine or distributed system in a model training platform on the network side, etc. This embodiment does not specifically limit this.

[0156] It is understandable that the application may be a local program (nativeApp) installed on the local terminal, or may be a webpage program (webApp) of a browser on the local terminal, which is not limited in this embodiment.

[0157] Optionally, in a possible implementation of this embodiment, the first encoding network includes a spatial feature encoder and a first position encoder, and the first encoding unit 602 can be used to use the spatial feature encoder to perform spatial feature encoding processing on the sample bird's-eye view feature map to obtain the result of spatial feature encoding processing of the sample bird's-eye view feature map; use the first position encoder to perform position encoding processing on the sample bird's-eye view feature map to obtain the result of position encoding processing of the sample bird's-eye view feature map; add the result of the spatial feature encoding processing and the result of the position encoding processing to obtain the bird's-eye view feature map encoding information.

[0158] Optionally, in a possible implementation of this embodiment, the second encoding network includes a driving state feature encoder and a second position encoder, and the second encoding unit 603 is used to use the driving state feature encoder to perform driving state feature encoding processing on the sample vehicle driving data to obtain the result of the driving state feature encoding processing of the sample vehicle driving data; use the second position encoder to perform position encoding processing on the sample vehicle driving data to obtain the result of the position encoding processing of the sample vehicle driving data; add the result of the driving state feature encoding processing and the result of the position encoding processing to obtain the vehicle driving feature coding information.

[0159] Optionally, in a possible implementation of this embodiment, the decoding network includes a decoder, a classification fully connected layer, and a regression fully connected layer. The first decoding unit 605 can be used to use the decoder to decode the fused feature encoding information to obtain a decoding result; input the decoding result into the classification fully connected layer to obtain the driving direction category information of the path planning; input the decoding result into the regression fully connected layer to obtain the trajectory information of the path planning; and obtain the path planning result based on the driving direction category information and the trajectory information.

[0160] In this embodiment, sample data can be acquired by a first acquisition unit, the sample data including sample vehicle driving data and a sample bird's-eye view feature map of the perception information corresponding to the sample vehicle driving data, and then the first encoding unit can encode the sample bird's-eye view feature map using the first encoding network to obtain bird's-eye view feature map encoding information, the second encoding unit can encode the sample vehicle driving data using the second encoding network to obtain vehicle driving feature encoding information, the first splicing unit can perform feature splicing on the bird's-eye view feature map encoding information and the vehicle driving feature encoding information to obtain fused feature encoding information, and the first decoding unit can decode the sample vehicle driving data using the second encoding network to obtain fused feature encoding information. The decoding network is used to decode the fused feature coding information to obtain a path planning result, so that the first training unit can update the path planning model to be trained based on the sample data and the path planning result to obtain a trained path planning model. Since the path planning model including the first coding network, the second coding network and the decoding network is trained based on the sample bird's-eye view feature map of the perception information corresponding to the sample vehicle driving data, a more accurate and effective end-to-end path planning model can be obtained, which ensures the performance and effectiveness of the path planning model, thereby ensuring the reliability and performance of the path planning of the autonomous driving vehicle.

[0161] Figure 7 A structural block diagram of a path planning device provided by an embodiment of the present application is shown as follows: Figure 7 As shown. The path planning device 700 of this embodiment may include an acquisition unit 701, a planning unit 702 and an acquisition unit 703. The acquisition unit 701 is used to acquire the vehicle driving data to be processed and the bird's-eye feature map of the perception information corresponding to the vehicle driving data; the planning unit 702 is used to input the vehicle driving data and the bird's-eye feature map into a pre-trained path planning model to obtain the result output by the path planning model; the acquisition unit 703 is used to obtain the path planning result based on the result output by the path planning model; wherein the path planning model is pre-trained using a training device for a path planning model, and the path planning model includes a first encoding network, a second encoding network and a decoding network.

[0162] It should be noted that part or all of the path planning device of this embodiment may be an application located in the local terminal, or may also be a functional unit such as a plug-in or software development kit (SDK) set in the application located in the local terminal, or may also be a processing engine located in a network-side server, or may also be a distributed system located on the network side, for example, a processing engine or distributed system in a path planning platform on the network side, etc. This embodiment does not specifically limit this.

[0163] It is understandable that the application may be a local program (nativeApp) installed on the local terminal, or may be a webpage program (webApp) of a browser on the local terminal, which is not limited in this embodiment.

[0164] In this embodiment, the vehicle driving data to be processed and the bird's-eye view feature map of the perception information corresponding to the vehicle driving data are obtained by the acquisition unit, and then the vehicle driving data and the bird's-eye view feature map can be input into the pre-trained path planning model by the planning unit to obtain the result output by the path planning model, so that the acquisition unit can obtain the path planning result based on the result output by the path planning model. The path planning model includes a first encoding network, a second encoding network and a decoding network. Since the pre-trained path planning model has better end-to-end path planning performance, the end-to-end path planning model is used to perform path planning on the acquired vehicle driving data and the bird's-eye view feature map of the perception information corresponding to the vehicle driving data. While ensuring the accuracy of the path planning result, the corresponding path planning result can be directly output based on the end-to-end planning model, which reduces computing power overhead, optimizes planning performance and efficiency, and thus improves the reliability of path planning of the autonomous driving vehicle.

[0165] In the technical solution of this application, the user personal information involved, such as the collection, storage, use, processing, transmission, provision and disclosure of user images and attribute data, etc., complies with the provisions of relevant laws and regulations and does not violate public order and good morals.

[0166] According to an embodiment of the present application, the present application also provides an electronic device, a readable storage medium and a computer program product.

[0167] According to an embodiment of the present application, further, an autonomous driving vehicle including the provided electronic device is provided, and the autonomous driving vehicle may include an unmanned vehicle of level L2 or above, for example, an unmanned commercial vehicle, an unmanned logistics vehicle, an autonomous driving logistics vehicle, etc.

[0168] Figure 8A schematic block diagram of an example electronic device 800 that can be used to implement an embodiment of the present application is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or required herein.

[0169] like Figure 8 As shown, the electronic device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the electronic device 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0170] Multiple components in the electronic device 800 are connected to the I / O interface 805, including: an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the electronic device 800 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0171] The computing unit 801 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 801 performs the various methods and processes described above, such as the method of path planning and the method of model training. For example, in some embodiments, the method of path planning and the method of model training may be implemented as a computer software program, which is tangibly included in a machine-readable medium, such as a storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the method of path planning and the method of model training described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to perform the path planning method and the model training method in any other appropriate manner (for example, by means of firmware).

[0172] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0173] The program code for implementing the method of the present application can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, implements the functions / operations specified in the flow chart and / or block diagram. The program code can be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0174] In the context of the present application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0175] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0176] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0177] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0178] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution disclosed in this application can be achieved, and this document is not limited here.

[0179] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A training method for a path planning model, characterized in that: The path planning model to be trained includes a first encoding network, a second encoding network and a decoding network, and the method includes: Acquire sample data, the sample data including sample vehicle driving data and a sample bird's-eye view feature map of perception information corresponding to the sample vehicle driving data; Using the first coding network, encoding the sample bird's-eye view feature image to obtain bird's-eye view feature image encoding information; Using the second coding network, encoding the sample vehicle driving data to obtain vehicle driving feature coding information; Performing feature splicing processing on the bird's-eye view feature image coding information and the vehicle driving feature coding information to obtain fused feature coding information; Using the decoding network, decoding the fused feature encoding information to obtain a path planning result; Based on the sample data and the path planning result, the path planning model to be trained is updated and trained to obtain a trained path planning model.

2. The method according to claim 1, characterized in that The first encoding network includes a spatial feature encoder and a first position encoder, and the first encoding network is used to encode the sample bird's-eye view feature map to obtain bird's-eye view feature map encoding information, including: Using the spatial feature encoder, performing spatial feature encoding processing on the sample bird's-eye view feature image to obtain a result of the spatial feature encoding processing of the sample bird's-eye view feature image; Using the first position encoder, performing position encoding processing on the sample bird's-eye view feature map to obtain a result of the position encoding processing of the sample bird's-eye view feature map; The result of the spatial feature encoding process and the result of the position encoding process are added together to obtain the bird's-eye view feature map encoding information.

3. The method according to claim 1, characterized in that The second coding network includes a driving state characteristic encoder and a second position encoder, and the second coding network is used to encode the sample vehicle driving data to obtain vehicle driving characteristic coding information, including: Using the driving state feature encoder, performing driving state feature encoding processing on the sample vehicle driving data to obtain a result of the driving state feature encoding processing of the sample vehicle driving data; Using the second position encoder, performing position encoding processing on the sample vehicle driving data to obtain a result of the position encoding processing of the sample vehicle driving data; The result of the driving state feature coding process and the result of the position coding process are added together to obtain the vehicle driving feature coding information.

4. The method according to claim 1, characterized in that: The decoding network includes a decoder, a classification fully connected layer, and a regression fully connected layer. The decoding network is used to decode the fusion feature encoding information to obtain a path planning result, including: Using the decoder, decoding the fused feature encoding information to obtain a decoding result; Inputting the result of the decoding process into the classification fully connected layer to obtain the driving direction category information of the path planning; Inputting the result of the decoding process into the regression fully connected layer to obtain trajectory information of the path planning; The path planning result is obtained based on the driving direction category information and the trajectory information.

5. A path planning method, characterized in that: The method comprises: Acquire vehicle driving data to be processed and a bird's-eye view feature map of perception information corresponding to the vehicle driving data; Inputting the vehicle driving data and the bird's-eye view feature map into a pre-trained path planning model to obtain a result output by the path planning model; Obtaining a path planning result based on a result output by the path planning model; The path planning model is pre-trained by the method according to any one of claims 1 to 4, and the path planning model includes a first encoding network, a second encoding network and a decoding network.

6. A training device for a path planning model, characterized in that: The path planning model to be trained includes a first encoding network, a second encoding network and a decoding network, and the device includes: A first acquisition unit is used to acquire sample data, wherein the sample data includes sample vehicle driving data and a sample bird's-eye view feature map of perception information corresponding to the sample vehicle driving data; A first encoding unit, configured to use the first encoding network to encode the sample bird's-eye view feature image to obtain bird's-eye view feature image encoding information; A second encoding unit is used to encode the sample vehicle driving data using the second encoding network to obtain vehicle driving feature encoding information; A first splicing unit is used to perform feature splicing processing on the bird's-eye view feature image coding information and the vehicle driving feature coding information to obtain fused feature coding information; A first decoding unit, configured to use the decoding network to decode the fused feature encoding information to obtain a path planning result; The first training unit is used to update the path planning model to be trained based on the sample data and the path planning result to obtain a trained path planning model.

7. A path planning device, characterized in that: The device comprises: An acquisition unit, used to acquire vehicle driving data to be processed and a bird's-eye view feature map of perception information corresponding to the vehicle driving data; A planning unit, used for inputting the vehicle driving data and the bird's-eye view feature map into a pre-trained path planning model to obtain a result output by the path planning model; An obtaining unit, configured to obtain a path planning result based on a result output by the path planning model; The path planning model is pre-trained using the device as described in claim 6, and the path planning model includes a first encoding network, a second encoding network and a decoding network.

8. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 5.

9. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-5.

10. An autonomous driving vehicle comprising the electronic device as claimed in claim 8.

Images