Vehicle track processing method and device and nonvolatile storage medium
By calculating the deviation of vehicle trajectory data and building a trajectory scoring model, the problem of low scoring efficiency and poor accuracy caused by manual intervention in vehicle trajectory planning is solved, and more efficient and accurate trajectory scoring is achieved.
Patent Information
- Application Number
- CN202310545354.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-15
- Publication Date
- 2024-11-15
AI Technical Summary
In the prior art, manual intervention is required when planning vehicle trajectory, resulting in low scoring efficiency, poor accuracy and high subjectivity.
By obtaining the vehicle's driving environment data and candidate trajectory data, the deviation between trajectories is calculated, the actual scoring results are generated, and the self-attention network model is used for training, and the trajectory scoring model is constructed to reduce manual intervention.
It realizes trajectory scoring without manual intervention, improves scoring efficiency and accuracy, and ensures that candidate trajectory scoring is more in line with the actual driving environment.
Smart Images

Figure CN118953381A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of autonomous driving, and in particular to a vehicle trajectory processing method, device and non-volatile storage medium. Background Art
[0002] An autonomous driving system is usually composed of multiple algorithm modules, such as a perception module for perceiving obstacles and a positioning module for locating the current position of the vehicle. The task of the trajectory planning module is to calculate the vehicle's driving trajectory in real time for several seconds in the future, and give the calculated result to the vehicle control module to control the vehicle to track the driving trajectory. The trajectory solution in the trajectory planning module is a complex computational problem. The problem includes two aspects: one is how to generate a trajectory given an intention, and the other is how to select the appropriate intention trajectory when multiple intention trajectories are obtained.
[0003] In the prior art, when selecting trajectories, it is mainly achieved through rule-based scoring methods. The specific steps are: first, design some evaluation and scoring functions for trajectories, such as the score of the smoothness of the trajectory, the score of the average forward speed of the trajectory, and the score of whether the trajectory changes lanes; second, score the trajectories of different intentions obtained using the above functions, and take a weighted average to find the total score of each trajectory; finally, sort the trajectories according to the score of each trajectory, and the trajectory with the highest score is the final selected trajectory. The above method mainly has the following problems: First, the evaluation and scoring function needs to be manually designed, and there are many different function design schemes. In addition, the more complex the scene, the more different functions are needed to evaluate; second, the weights of the weighted average of different trajectory scores require manual participation, which leads to a high degree of subjectivity in the scoring results and easily leads to inaccurate trajectory selection. Summary of the invention
[0004] The embodiments of the present invention provide a vehicle trajectory processing method, device and non-volatile storage medium to at least solve the technical problem in the related art that manual intervention is required when selecting a vehicle's planned trajectory, which has a large subjective influence and leads to low scoring efficiency and poor accuracy.
[0005] According to one aspect of an embodiment of the present invention, a vehicle trajectory processing method is provided, comprising: acquiring first driving environment data of the vehicle, multiple groups of first candidate trajectory data and first actual trajectory data corresponding to the first driving environment data; calculating deviations between the multiple groups of first candidate trajectory data and the first actual trajectory data in the same time period, and obtaining actual scoring results corresponding to the multiple groups of first candidate trajectory data; determining a training result according to the first driving environment data, the multiple groups of first candidate trajectory data, and the actual scoring results corresponding to the multiple groups of first candidate trajectory data.
[0006] According to another aspect of an embodiment of the present invention, a vehicle trajectory processing device is also provided, including: an acquisition module, used to acquire first driving environment data of the vehicle, multiple groups of first candidate trajectory data and first actual trajectory data corresponding to the above-mentioned first driving environment data; a calculation module, used to calculate the deviations between the above-mentioned multiple groups of first candidate trajectory data and the above-mentioned first actual trajectory data in the same time period, and obtain actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data; a determination module, used to determine the training result according to the above-mentioned first driving environment data, the above-mentioned multiple groups of first candidate trajectory data, and the actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data.
[0007] According to another aspect of an embodiment of the present invention, a non-volatile storage medium is further provided, wherein the non-volatile storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing any one of the above-mentioned vehicle trajectory processing methods.
[0008] In an embodiment of the present invention, by obtaining first driving environment data of a vehicle, multiple groups of first candidate trajectory data and first actual trajectory data corresponding to the above-mentioned first driving environment data; calculating the deviations between the above-mentioned multiple groups of first candidate trajectory data and the above-mentioned first actual trajectory data in the same time period, and obtaining the actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data; determining the training results according to the above-mentioned first driving environment data, the above-mentioned multiple groups of first candidate trajectory data, and the actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data, the purpose of accurately obtaining the scoring values of multiple candidate trajectories without manual intervention is achieved, thereby achieving the technical effect of improving the vehicle's trajectory scoring efficiency, reducing manual access, and thus improving the scoring accuracy, thereby solving the technical problem in the related art that manual intervention is required when selecting the planned trajectory of the vehicle, which has a large subjective influence, resulting in low scoring efficiency and poor accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0009] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:
[0010] Figure 1 is a schematic diagram of a vehicle trajectory processing method according to an embodiment of the present invention;
[0011] Figure 2 is an optional distribution diagram of historical actual trajectories and historical candidate trajectories according to an embodiment of the present invention;
[0012] Figure 3 is a schematic diagram of a vehicle trajectory processing device according to an embodiment of the present invention. DETAILED DESCRIPTION
[0013] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.
[0014] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or modules is not necessarily limited to those steps or modules that are clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices.
[0015] According to an embodiment of the present invention, a method embodiment of vehicle trajectory processing is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.
[0016] Figure 1 is a flow chart of a vehicle trajectory processing method according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:
[0017] Step S102 , obtaining first driving environment data of the vehicle, and a plurality of sets of first candidate trajectory data and first actual trajectory data corresponding to the first driving environment data.
[0018] Optionally, the first driving environment data is obtained based on the historical driving behavior of the vehicle, and is the historical driving environment data of the vehicle; the multiple sets of first candidate trajectory data are multiple sets of historical candidate trajectory data corresponding to the historical driving environment data of the vehicle; and the first actual trajectory data are historical actual trajectory data corresponding to the historical driving environment data of the vehicle. For example, it can be the historical actual driving trajectory of the target object (such as the driver). Among the multiple sets of first candidate trajectory data, each set of first candidate trajectory data corresponds to a historical candidate trajectory.
[0019] Optionally, the first driving environment data corresponds to multiple sets of first candidate trajectory data and first actual trajectory data, which can be understood as each frame of driving data including driving environment data, and corresponding multiple candidate trajectory data and actual trajectory data.
[0020] Optionally, the first driving environment data at least includes: current position information of the vehicle, current movement information of other vehicles other than the vehicle within a preset first distance range, and current lane information, wherein the current lane information includes at least one of the following: number of lanes within the first distance range, lane line information, zebra crossing information, road edge information, and road speed limit information. The multiple sets of first driving trajectory data correspond to different driving intentions, and the multiple driving intentions may include but are not limited to: going straight, turning left, turning right, changing lanes, and the like.
[0021] Step S104 , calculating deviations between the plurality of sets of first candidate trajectory data and the first actual trajectory data in the same time period, and obtaining actual scoring results corresponding to the plurality of sets of first candidate trajectory data.
[0022] Optionally, the deviation obtained by the above calculation is used to indicate the degree of deviation between the candidate trajectory and the actual trajectory. The greater the degree of deviation, the worse the effect of the candidate trajectory. Based on this, by calculating the deviation between the above multiple groups of first candidate trajectory data and the above first actual trajectory data in the same time period, the actual score value of the candidate trajectory is determined based on the degree of deviation between the candidate trajectory and the actual trajectory, eliminating the interference caused by human subjective factors when determining the score value through manual intervention, so that the actual score value obtained is more in line with reality.
[0023] Optionally, taking the above-mentioned multiple groups of first candidate trajectory data corresponding to multiple historical candidate trajectories and the first actual trajectory data corresponding to the historical actual trajectory as an example, during the driving process, the vehicle performs trajectory planning at a certain interval (such as 0.5 seconds), and each planned path is a driving trajectory within a predetermined time (such as 10 seconds). Figure 2 is an optional schematic diagram of the distribution of historical actual trajectories and historical candidate trajectories according to an embodiment of the present invention, such as Figure 2 As shown in (a), at a certain moment of 0.5 seconds, the vehicle plans multiple candidate trajectories with different driving intentions from 0.5 seconds to 10.5 seconds. At this time, the deviation between the vehicle's multiple historical candidate trajectories from 0.5 seconds to 10.5 seconds and the historical actual trajectory can be calculated; at 1 second, the vehicle will re-plan the path and plan multiple candidate trajectories with different driving intentions from 1 second to 11 seconds, and so on.
[0024] In an optional embodiment, the above-mentioned calculating the deviations between the above-mentioned multiple groups of first candidate trajectory data and the above-mentioned first actual trajectory data in the same time period to obtain the actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data respectively includes: calculating the average distance deviations between the above-mentioned multiple groups of first candidate trajectory data and the above-mentioned first actual trajectory data in the same time period; based on the above-mentioned average distance deviation, determining the actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data respectively.
[0025] Optionally, the actual scoring values corresponding to the multiple sets of first candidate trajectory data are determined by calculating the average distance deviation between the multiple sets of first candidate trajectory data and the first actual trajectory data in the same time period. The larger the average distance deviation, the smaller the actual scoring value of the corresponding candidate trajectory. In the above manner, the actual scoring value of the candidate trajectory is determined based on the average distance deviation of the multiple sets of first candidate trajectory data relative to the first actual trajectory data, which eliminates the interference caused by subjective factors when the scoring value is determined by manual intervention, so that the actual scoring value obtained is more accurate and reliable.
[0026] In an optional embodiment, the average distance deviation between any one group of candidate trajectory data and the first actual trajectory data in the above-mentioned multiple groups of first candidate trajectory data in the above-mentioned same time period is calculated in the following manner, including: determining multiple sampling moments in the above-mentioned same time period; calculating the distance deviation between the above-mentioned any one group of candidate trajectory data and the above-mentioned first actual trajectory data at each sampling moment in the above-mentioned multiple sampling moments to obtain multiple distance deviations, wherein the time interval between any two adjacent sampling moments in the above-mentioned same time period is a predetermined time interval; calculating the average value of the above-mentioned multiple distance deviations to obtain the above-mentioned average distance deviation between the above-mentioned any one group of candidate trajectory data and the above-mentioned first actual trajectory data.
[0027] Optionally, multiple sampling moments within the same time period are determined according to a predetermined time interval (such as 0.1 seconds). For example, for a frame of historical driving data collected at 1 second, including historical environmental data, and corresponding multiple sets of historical candidate trajectory data (corresponding to multiple sets of historical candidate trajectories) and historical actual trajectory data (corresponding to historical actual trajectories), the corresponding time period is from 1 second to 11 seconds, such as Figure 2 As shown in (b), from 1 second to 11 seconds of the historical candidate trajectory and trajectory segment, every 0.1 second is used as a sampling moment, and multiple sampling moments are obtained. The distance deviations of the historical candidate trajectories and the historical actual trajectories at each sampling moment are determined, and the distance deviations corresponding to the multiple sampling moments are obtained. The average of the distance deviations corresponding to the multiple sampling moments is calculated to obtain the average distance deviation. It can be understood that the average deviation can be used to characterize the average deviation degree between the candidate trajectory and the actual trajectory.
[0028] Step S106 , determining a training result according to the first driving environment data, the multiple sets of first candidate trajectory data, and actual scoring results respectively corresponding to the multiple sets of first candidate trajectory data.
[0029] Optionally, the training result is determined based on the first driving environment data, the multiple groups of first candidate trajectory data, and the actual scoring results corresponding to the multiple groups of first candidate trajectory data, specifically including: based on the first driving environment data, the multiple groups of first candidate trajectory data, and the actual scoring results corresponding to the multiple groups of first candidate trajectory data, a self-attention network model is used for training to obtain the training result.
[0030] Optionally, the above training results can be training parameters output by training, or can be a trained trajectory scoring model; the above training parameters can be used to construct a scoring prediction model, which is used for subsequent trajectory scoring prediction based on the constructed trajectory scoring model; the above scoring prediction model can be directly used in the trajectory scoring model to perform trajectory scoring prediction, thereby improving the efficiency of trajectory scoring prediction.
[0031] In an optional embodiment, the training result is determined based on the first driving environment data, the multiple groups of first candidate trajectory data, and the actual scoring results corresponding to the multiple groups of first candidate trajectory data, including: performing fusion training on the first driving environment data and the multiple groups of first candidate trajectory data to obtain the predicted scoring results corresponding to the multiple groups of first candidate trajectory data; obtaining the training result based on the predicted scoring results and the actual scoring results corresponding to the multiple groups of first candidate trajectory data.
[0032] Optionally, during the training process, the first driving environment data and the multiple sets of first candidate trajectory data are integrated for training, thereby obtaining prediction scoring results corresponding to the multiple sets of first candidate trajectory data; the purpose of the training is to make the prediction scoring results finally output by the model as close as possible to the actual scoring results. By comparing and judging the prediction scoring results and the actual scoring results, the smaller the difference between the prediction scoring results and the actual scoring results, the better the effect of the corresponding model training results.
[0033] In an optional embodiment, the above-mentioned first driving environment data and the above-mentioned multiple groups of first candidate trajectory data are fused to obtain a fusion result, including: performing a first encoding process on the above-mentioned first driving environment data to obtain an environment code; performing a second encoding process on the above-mentioned multiple groups of first candidate trajectory data to obtain multiple trajectory codes; performing the above-mentioned fusion training based on the above-mentioned multiple trajectory codes and the above-mentioned environment codes respectively, to obtain prediction scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data respectively.
[0034] Optionally, the first driving environment data and the multiple groups of candidate trajectory data are encoded respectively to obtain corresponding environment codes and multiple groups of trajectory codes. The environment codes and multiple groups of trajectory codes are continuously fused and learned, and finally the prediction and scoring results corresponding to the multiple groups of first candidate trajectory data are obtained. The above method can be understood as network fusion, that is, an environment network is constructed based on the first driving environment data, and a trajectory network is constructed based on multiple groups of first candidate trajectory data, and the constructed trajectory network and environment network are fused and learned, and finally the prediction and scoring results corresponding to the multiple groups of first candidate trajectory data are obtained. In the above manner, the training results can learn the relationship between driving environment information and trajectory scores. The training results finally output can be used for subsequent scoring prediction of multiple candidate trajectories based on driving environment information.
[0035] In an optional embodiment, when the above-mentioned training result is a trajectory scoring model, after determining the training result based on the above-mentioned first driving environment data, the above-mentioned multiple groups of first candidate trajectory data, and the actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data, the above-mentioned method also includes: obtaining the second driving environment data of the above-mentioned vehicle and the multiple groups of second candidate trajectory data corresponding to the above-mentioned second driving environment data; based on the above-mentioned second driving environment data and the above-mentioned multiple groups of second candidate trajectory data, using the above-mentioned trajectory scoring model, obtain the predicted scoring results corresponding to the above-mentioned multiple groups of second candidate trajectory data.
[0036] Optionally, the second driving environment data of the above-mentioned vehicle, the current driving environment data collected in real time during the vehicle's driving, and the corresponding multiple sets of second candidate trajectory data can be multiple sets of current candidate trajectory data corresponding to the current driving environment data. In the case where the above-mentioned training result is a trajectory scoring model, the second driving environment data of the vehicle and the corresponding multiple sets of second candidate trajectory data can be directly input into the pre-trained trajectory scoring model to quickly obtain the prediction scoring results corresponding to the multiple sets of second candidate trajectories. It can be understood that the prediction scoring result can reflect the adaptability of the candidate trajectory to the current driving environment. The higher the candidate trajectory prediction score value, the more suitable it is for the current driving environment.
[0037] In an optional embodiment, when the above-mentioned training results are model parameters, after determining the training results based on the above-mentioned first driving environment data, the above-mentioned multiple groups of first candidate trajectory data, and the actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data, the above-mentioned method also includes: obtaining the second driving environment data of the above-mentioned vehicle and the multiple groups of second candidate trajectory data corresponding to the above-mentioned second driving environment data; based on the above-mentioned second driving environment data and the above-mentioned multiple groups of second candidate trajectory data, using the above-mentioned trajectory scoring model, obtain the predicted scoring results corresponding to the above-mentioned multiple groups of second candidate trajectory data.
[0038] Optionally, when the above training results are model parameters, a trajectory scoring model can be first constructed based on the model parameters. The second driving environment data of the vehicle and the corresponding multiple sets of second candidate trajectory data are input into the pre-trained trajectory scoring model, and the prediction scoring results corresponding to the multiple sets of second candidate trajectories can be quickly obtained. It can be understood that the prediction scoring results can reflect the adaptability of the candidate trajectory to the current driving environment. The higher the candidate trajectory prediction score value, the more suitable it is for the current driving environment.
[0039] In an optional embodiment, after obtaining the predicted scoring results corresponding to the above-mentioned multiple groups of second candidate trajectory data respectively based on the above-mentioned second driving environment data and the above-mentioned multiple groups of second candidate trajectory data using the above-mentioned trajectory scoring model, the above-mentioned method also includes: determining the candidate trajectory data with the highest predicted scoring result among the above-mentioned multiple groups of second candidate trajectory data; and using the driving trajectory corresponding to the candidate trajectory data with the highest predicted scoring result as the target driving trajectory of the above-mentioned vehicle.
[0040] It can be understood that since the above-mentioned trajectory scoring model is trained based on the driving environment and the deviation between the candidate trajectory and the actual trajectory, the above-mentioned scoring prediction results can reflect the degree of adaptability of the candidate trajectory to the current driving environment. The higher the score value, the better the effect of the candidate trajectory. When automatically determining the driving vehicle, the candidate trajectory with the highest score value is used as the target running trajectory. The target running trajectory obtained in this way is more in line with the needs of the current driving environment.
[0041] Through the above steps S102 to S106, the purpose of accurately obtaining the scoring values of multiple candidate trajectories without manual intervention can be achieved, thereby achieving the technical effect of improving the vehicle trajectory scoring efficiency, reducing manual intervention, and thus improving the scoring accuracy, thereby solving the technical problem in the related technology that manual intervention is required when selecting the planned trajectory of the vehicle, which has a large subjective influence, resulting in low scoring efficiency and poor accuracy.
[0042] Based on the above embodiments and optional embodiments, the present invention proposes a flow chart of an optional vehicle trajectory processing method, the method comprising:
[0043] Step S1, obtaining historical driving environment data of the vehicle, multiple groups of historical candidate trajectory data corresponding to the historical driving environment data, and historical actual trajectory data corresponding to the historical driving environment data.
[0044] Step S2, calculating the deviations between multiple sets of historical candidate trajectory data and historical actual trajectory data in the same period, and obtaining actual scoring results corresponding to the multiple sets of historical candidate trajectory data, specifically includes the following sub-steps:
[0045] Step S21, with a time interval of 0.1 seconds, multiple sampling moments are determined from the same time period, and the distance deviation between each set of historical candidate trajectory data and historical actual trajectory data at each sampling moment is calculated, that is, the distance deviation between the candidate trajectory data and the historical actual trajectory data is calculated every 0.1 seconds in the same time period, and the distance deviations corresponding to the multiple sampling moments are obtained;
[0046] Step S22, calculating the average values of the distance deviations corresponding to the multiple sampling moments, and obtaining the average distance deviations corresponding to the multiple groups of historical candidate trajectory data;
[0047] Step S23: determining actual scoring results corresponding to the multiple sets of historical candidate trajectory data based on the average distance deviations corresponding to the multiple sets of historical candidate trajectory data.
[0048] Step S3, determining the training result according to the historical driving environment data, the multiple sets of historical candidate trajectory data, and the actual scoring results corresponding to the multiple sets of historical candidate trajectory data, specifically includes the following sub-steps:
[0049] Step S31, performing a first encoding process on the historical driving environment data to obtain an environment code;
[0050] Step S32, performing a second encoding process on the multiple groups of historical candidate trajectory data to obtain multiple trajectory codes;
[0051] Step S33, performing fusion training based on multiple trajectory codes and environment codes respectively, to obtain prediction scoring results corresponding to multiple groups of historical candidate trajectory data respectively;
[0052] Step S34: determining a trajectory scoring model based on the predicted scoring results and the actual scoring results corresponding to the multiple groups of historical candidate trajectory data.
[0053] Step S4, obtaining current driving environment data of the vehicle at the current driving moment, and multiple sets of current candidate trajectory data corresponding to the current driving environment data.
[0054] Step S5, based on the current driving environment data and multiple sets of current candidate trajectory data, the above-mentioned trained trajectory scoring model is used to obtain prediction scoring results corresponding to the multiple sets of current candidate trajectory data.
[0055] Step S6, determining the candidate trajectory data with the highest prediction score among the multiple sets of current candidate trajectory data; and using the driving trajectory corresponding to the candidate trajectory data with the highest prediction score as the target driving trajectory of the vehicle.
[0056] It should be noted that the embodiments of the present invention can at least achieve the following technical effects: it does not require a large amount of manual work to design the trajectory scoring function; it can make the final trajectory more in line with the actual driving environment and more like the choice of a human driver in a data-driven manner; it improves the efficiency of vehicle trajectory scoring, reduces manual intervention, and thus improves scoring accuracy.
[0057] In this embodiment, a vehicle trajectory processing device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the terms "module" and "device" can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.
[0058] According to an embodiment of the present invention, there is also provided an embodiment of a device for implementing the above-mentioned vehicle trajectory processing method. Figure 3 is a schematic diagram of the structure of a vehicle trajectory processing device according to an embodiment of the present invention, such as Figure 3 As shown, the above-mentioned vehicle trajectory processing device includes: an acquisition module 300, a calculation module 302, and a determination module 304, wherein:
[0059] The acquisition module 300 is used to acquire first driving environment data of the vehicle, and a plurality of sets of first candidate trajectory data and first actual trajectory data corresponding to the first driving environment data;
[0060] The calculation module 302 is connected to the acquisition module 300 and is used to calculate the deviations between the plurality of sets of first candidate trajectory data and the first actual trajectory data in the same period, and obtain the actual scoring results corresponding to the plurality of sets of first candidate trajectory data;
[0061] The determination module 304 is connected to the calculation module 302 and is used to determine the training result according to the first driving environment data, the multiple sets of first candidate trajectory data, and the actual scoring results respectively corresponding to the multiple sets of first candidate trajectory data.
[0062] In an embodiment of the present invention, the acquisition module 300 is provided to acquire the first driving environment data of the vehicle, the multiple groups of first candidate trajectory data and the first actual trajectory data corresponding to the first driving environment data; the calculation module 302 is connected to the acquisition module 300 to calculate the deviations between the multiple groups of first candidate trajectory data and the first actual trajectory data in the same period, and obtain the actual scoring results corresponding to the multiple groups of first candidate trajectory data; the determination module 304 is connected to the calculation module 302 to determine the training result according to the first driving environment data, the multiple groups of first candidate trajectory data, and the actual scoring results corresponding to the multiple groups of first candidate trajectory data, so as to achieve the purpose of accurately acquiring the scoring values of multiple candidate trajectories without manual intervention, thereby achieving the technical effect of improving the trajectory scoring efficiency of the vehicle, reducing manual access, and thus improving the scoring accuracy, thereby solving the technical problem in the related art that manual intervention is required when selecting the planned trajectory of the vehicle, which has a large subjective influence, resulting in low scoring efficiency and poor accuracy.
[0063] According to an embodiment of the present invention, another vehicle trajectory processing device is also provided, which includes an environment perception module, a trajectory planning module, and a planning decision module, wherein: the above-mentioned environment perception module is used to obtain the current driving environment information of the above-mentioned vehicle; the above-mentioned trajectory planning module is used to determine multiple current candidate trajectories of the above-mentioned vehicle, wherein the above-mentioned multiple current candidate trajectories are planned trajectories of the above-mentioned vehicle within a reserved time period starting from the current moment under multiple different driving intentions; the above-mentioned planning decision module is connected to the above-mentioned environment perception module and the above-mentioned trajectory planning module, and is used to obtain the scoring value trajectories corresponding to the above-mentioned multiple current candidate trajectories respectively based on the above-mentioned current driving environment information and the above-mentioned multiple current candidate trajectories, using a trajectory scoring model.
[0064] Optionally, the above-mentioned environmental perception module may include but is not limited to a map module, a positioning module, an image acquisition module, a perception module, etc., which is used to obtain the current position of the vehicle, the current movement information of other vehicles, the current lane information (such as the number of lanes near the vehicle, lane line information, zebra crossing information, road edge information, and road speed limit information) and other current driving environment information.
[0065] In an optional embodiment, the above-mentioned planning decision module includes a scoring submodule and a decision submodule, wherein the above-mentioned scoring submodule is used to obtain the scoring values corresponding to the above-mentioned multiple current candidate trajectories respectively by using a trajectory scoring model based on the above-mentioned current driving environment information and the above-mentioned multiple current candidate trajectories; the above-mentioned decision submodule is used to use the current candidate trajectory with the highest scoring value among the above-mentioned multiple current candidate trajectories as the target running trajectory of the above-mentioned vehicle. In the above manner, after obtaining the scoring values corresponding to the multiple current candidate trajectories respectively, the driving trajectory can be further determined according to the scoring values, that is, the current candidate trajectory with the highest scoring value is used as the target running trajectory of the above-mentioned vehicle, thereby achieving the purpose of performing optimal trajectory planning.
[0066] It should be noted that the above modules can be implemented by software or hardware. For example, for the latter, it can be implemented in the following ways: the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.
[0067] It should be noted that the above-mentioned environment perception module 300, trajectory planning module 302, and planning decision module 304 correspond to steps S102 to S106 in the embodiment, and the examples and application scenarios implemented by the above-mentioned modules and corresponding steps are the same, but are not limited to the contents disclosed in the above-mentioned embodiment. It should be noted that the above-mentioned modules can be run in a computer terminal as part of the device.
[0068] It should be noted that the optional or preferred implementation of this embodiment can refer to the relevant description in the embodiment, which will not be repeated here.
[0069] The above-mentioned vehicle trajectory processing device may also include a processor and a memory. The above-mentioned environment perception module 300, trajectory planning module 302, planning decision module 304, etc. are all stored in the memory as program modules, and the processor executes the above-mentioned program modules stored in the memory to realize corresponding functions.
[0070] The processor includes a kernel, which retrieves the corresponding program module from the memory. The kernel may be one or more. The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. The memory includes at least one memory chip.
[0071] According to an embodiment of the present application, an embodiment of a non-volatile storage medium is also provided. Optionally, in this embodiment, the non-volatile storage medium includes a stored program, wherein when the program is executed, the device where the non-volatile storage medium is located is controlled to execute any of the above-mentioned vehicle trajectory processing methods.
[0072] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group, and the non-volatile storage medium includes a stored program.
[0073] Optionally, when the program is running, the device where the non-volatile storage medium is located is controlled to perform the following functions: obtaining the first driving environment data of the vehicle, multiple groups of first candidate trajectory data and first actual trajectory data corresponding to the above-mentioned first driving environment data; calculating the deviations between the above-mentioned multiple groups of first candidate trajectory data and the above-mentioned first actual trajectory data in the same time period, and obtaining actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data; determining the training result according to the above-mentioned first driving environment data, the above-mentioned multiple groups of first candidate trajectory data, and the actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data.
[0074] According to an embodiment of the present application, an embodiment of a processor is also provided. Optionally, in this embodiment, the processor is used to run a program, wherein when the program is run, any one of the vehicle trajectory processing methods is executed.
[0075] According to an embodiment of the present application, an embodiment of a computer program product is also provided. When executed on a data processing device, the program is suitable for executing the steps of initializing any one of the above-mentioned vehicle trajectory processing methods.
[0076] Optionally, the above-mentioned computer program product, when executed on a data processing device, is suitable for executing a program initialized with the following method steps: obtaining first driving environment data of the vehicle, multiple groups of first candidate trajectory data and first actual trajectory data corresponding to the above-mentioned first driving environment data; calculating the deviations between the above-mentioned multiple groups of first candidate trajectory data and the above-mentioned first actual trajectory data in the same time period, and obtaining actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data; determining the training result according to the above-mentioned first driving environment data, the above-mentioned multiple groups of first candidate trajectory data, and the actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data.
[0077] Optionally, the above-mentioned computer program product, when executed on a data processing device, is suitable for executing a program that is initialized with the following method steps: calculating the average distance deviations between the above-mentioned multiple groups of first candidate trajectory data and the above-mentioned first actual trajectory data in the above-mentioned same time period; based on the above-mentioned average distance deviations, determining the actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data respectively.
[0078] Optionally, the above-mentioned computer program product, when executed on a data processing device, is suitable for executing a program initialized with the following method steps: determining multiple sampling moments within the above-mentioned same time period; calculating the distance deviation between the above-mentioned any one group of candidate trajectory data and the above-mentioned first actual trajectory data at each of the above-mentioned multiple sampling moments to obtain multiple distance deviations, wherein the time interval between any two adjacent sampling moments in the above-mentioned same time period is a predetermined time interval; calculating the average value of the above-mentioned multiple distance deviations to obtain the above-mentioned average distance deviation between the above-mentioned any one group of candidate trajectory data and the above-mentioned first actual trajectory data.
[0079] Optionally, the above-mentioned computer program product, when executed on a data processing device, is suitable for executing a program initialized with the following method steps: performing fusion training on the above-mentioned first driving environment data and the above-mentioned multiple groups of first candidate trajectory data to obtain the predicted scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data respectively; and obtaining the above-mentioned training results based on the predicted scoring results and actual scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data respectively.
[0080] Optionally, the above-mentioned computer program product, when executed on a data processing device, is suitable for executing a program initialized with the following method steps: performing a first encoding process on the above-mentioned first driving environment data to obtain an environment code; performing a second encoding process on the above-mentioned multiple groups of first candidate trajectory data to obtain multiple trajectory codes; performing the above-mentioned fusion training based on the above-mentioned multiple trajectory codes and the above-mentioned environment codes, respectively, to obtain prediction scoring results corresponding to the above-mentioned multiple groups of first candidate trajectory data.
[0081] Optionally, the above-mentioned computer program product, when executed on a data processing device, is suitable for executing a program that initializes the following method steps: obtaining the second driving environment data of the above-mentioned vehicle, and multiple groups of second candidate trajectory data corresponding to the above-mentioned second driving environment data; based on the above-mentioned second driving environment data and the above-mentioned multiple groups of second candidate trajectory data, using the above-mentioned trajectory scoring model, obtaining the predicted scoring results corresponding to the above-mentioned multiple groups of second candidate trajectory data respectively.
[0082] Optionally, the above-mentioned computer program product, when executed on a data processing device, is suitable for executing a program that initializes the following method steps: obtaining the second driving environment data of the above-mentioned vehicle, and multiple groups of second candidate trajectory data corresponding to the above-mentioned second driving environment data; based on the above-mentioned second driving environment data and the above-mentioned multiple groups of second candidate trajectory data, using the above-mentioned trajectory scoring model, obtaining the predicted scoring results corresponding to the above-mentioned multiple groups of second candidate trajectory data respectively.
[0083] Optionally, the above-mentioned computer program product, when executed on a data processing device, is suitable for executing a program initialized with the following method steps: determining the candidate trajectory data with the highest predicted score result among the above-mentioned multiple groups of second candidate trajectory data; and using the driving trajectory corresponding to the candidate trajectory data with the highest predicted score result as the target driving trajectory of the above-mentioned vehicle.
[0084] An embodiment of the present invention provides an electronic device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, the following steps are implemented: obtaining first driving environment data of a vehicle, multiple groups of first candidate trajectory data and first actual trajectory data corresponding to the first driving environment data; calculating deviations between the multiple groups of first candidate trajectory data and the first actual trajectory data in the same time period, respectively, to obtain actual scoring results corresponding to the multiple groups of first candidate trajectory data; determining a training result based on the first driving environment data, the multiple groups of first candidate trajectory data, and the actual scoring results corresponding to the multiple groups of first candidate trajectory data.
[0085] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.
[0086] In the above embodiments of the present invention, 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.
[0087] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the above modules can be a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, modules or indirect coupling or communication connection of modules, which can be electrical or other forms.
[0088] The modules described above as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment.
[0089] In addition, each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of software functional modules.
[0090] If the above-mentioned integrated module is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable non-volatile storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a non-volatile storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the various embodiments of the present invention. The aforementioned non-volatile storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.
[0091] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A vehicle trajectory processing method, characterized in that: include: Acquire first driving environment data of the vehicle, and a plurality of sets of first candidate trajectory data and first actual trajectory data corresponding to the first driving environment data; Calculating deviations between the multiple sets of first candidate trajectory data and the first actual trajectory data in the same time period, and obtaining actual scoring results corresponding to the multiple sets of first candidate trajectory data; A training result is determined according to the first driving environment data, the multiple groups of first candidate trajectory data, and actual scoring results respectively corresponding to the multiple groups of first candidate trajectory data.
2. The method according to claim 1, characterized in that The calculating deviations between the plurality of groups of first candidate trajectory data and the first actual trajectory data in the same time period to obtain actual scoring results corresponding to the plurality of groups of first candidate trajectory data respectively includes: Calculating average distance deviations between the plurality of groups of first candidate trajectory data and the first actual trajectory data in the same time period; Based on the average distance deviation, actual scoring results respectively corresponding to the multiple groups of first candidate trajectory data are determined.
3. The method according to claim 2, characterized in that The average distance deviation between any one set of candidate trajectory data and the first actual trajectory data in the plurality of sets of first candidate trajectory data in the same time period is calculated in the following manner, including: Determine a plurality of sampling moments within the same period; Calculating the distance deviation between the any group of candidate trajectory data and the first actual trajectory data at each sampling moment in the multiple sampling moments to obtain multiple distance deviations, wherein the time interval between any two adjacent sampling moments in the same time period is a predetermined time interval; An average value of the multiple distance deviations is calculated to obtain the average distance deviation between the arbitrary set of candidate trajectory data and the first actual trajectory data.
4. The method according to claim 1, characterized in that: The determining of the training result according to the first driving environment data, the plurality of sets of first candidate trajectory data, and actual scoring results respectively corresponding to the plurality of sets of first candidate trajectory data includes: Performing fusion training on the first driving environment data and the multiple groups of first candidate trajectory data to obtain prediction scoring results corresponding to the multiple groups of first candidate trajectory data respectively; The training result is obtained based on the predicted scoring results and the actual scoring results respectively corresponding to the multiple groups of first candidate trajectory data.
5. The method according to claim 4, characterized in that The fusing the first driving environment data and the plurality of sets of first candidate trajectory data to obtain a fusion result includes: Performing a first encoding process on the first driving environment data to obtain an environment code; Performing a second encoding process on the plurality of groups of first candidate trajectory data to obtain a plurality of trajectory codes; The fusion training is performed based on the multiple trajectory codes and the environment code to obtain prediction score results corresponding to the multiple groups of first candidate trajectory data.
6. The method according to claim 1, characterized in that In the case where the training result is a trajectory scoring model, after determining the training result according to the first driving environment data, the multiple groups of first candidate trajectory data, and the actual scoring results respectively corresponding to the multiple groups of first candidate trajectory data, the method further includes: Acquire second driving environment data of the vehicle, and multiple groups of second candidate trajectory data corresponding to the second driving environment data; Based on the second driving environment data and the multiple groups of second candidate trajectory data, the trajectory scoring model is used to obtain prediction scoring results corresponding to the multiple groups of second candidate trajectory data respectively.
7. The method according to claim 6, characterized in that In the case where the training result is a model parameter, after determining the training result according to the first driving environment data, the multiple sets of first candidate trajectory data, and the actual scoring results respectively corresponding to the multiple sets of first candidate trajectory data, the method further includes: Acquire second driving environment data of the vehicle, and multiple groups of second candidate trajectory data corresponding to the second driving environment data; Based on the second driving environment data and the multiple groups of second candidate trajectory data, the trajectory scoring model is used to obtain prediction scoring results corresponding to the multiple groups of second candidate trajectory data respectively.
8. The method according to claim 6 or 7, characterized in that: After obtaining prediction scoring results respectively corresponding to the multiple sets of second candidate trajectory data by using the trajectory scoring model based on the second driving environment data and the multiple sets of second candidate trajectory data, the method further includes: Determine, among the plurality of sets of second candidate trajectory data, candidate trajectory data with the highest prediction score result; The driving trajectory corresponding to the candidate trajectory data with the highest prediction score result is used as the target driving trajectory of the vehicle.
9. A vehicle trajectory processing device, characterized in that: include: An acquisition module, configured to acquire first driving environment data of the vehicle, and a plurality of groups of first candidate trajectory data and first actual trajectory data corresponding to the first driving environment data; a calculation module, configured to calculate deviations between the plurality of sets of first candidate trajectory data and the first actual trajectory data in the same time period, and obtain actual scoring results corresponding to the plurality of sets of first candidate trajectory data; A determination module is used to determine a training result according to the first driving environment data, the multiple groups of first candidate trajectory data, and actual scoring results respectively corresponding to the multiple groups of first candidate trajectory data.
10. A non-volatile storage medium, characterized in that: The non-volatile storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executed by the vehicle trajectory processing method described in any one of claims 1 to 8.
