Automatic driving scene data processing method and device and storage medium

Abstract

The embodiment of the invention provides an automatic driving scene data processing method and device and a storage medium. At least one traffic track on the road section can be identified according to dynamic traffic data in the roadside sensing data; carrying out key frame extraction on the at least one traffic track to obtain at least one group of key frames; performing decision planning according to the at least one group of key frames and a preset constraint condition to obtain at least one driving track; and determining automatic driving scene data according to static road network information in the roadside sensing data and the at least one driving track. Therefore, the automatic driving scene data in the rich scene can be obtained without actual measurement of the automatic driving vehicle on the road. Besides, variability and randomness of the automatic driving scene data can be increased by utilizing settable characteristics of constraint conditions, so that driving tracks in various different scenes are derived, and the coverage rate of the automatic driving scene data to different automatic driving scenes is increased.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to a method, device and storage medium for processing autonomous driving scene data. Background Art

[0002] Autonomous driving technology includes a large number of machine learning and deep learning algorithms, and the training and verification of the above algorithms involve a large amount of test data. When the algorithm is released or updated, in order to reduce the cost and time of on-road testing, the Autonomous Driving Senario (ADS) is usually used to simulate and regress the algorithm.

[0003] The autonomous driving scenario library contains multiple test cases. When constructing the scenario library, test cases can be obtained by combining multiple dimensional data such as "road section attributes, main vehicle behavior, surrounding vehicle positions, surrounding vehicle behavior", etc. Based on this combination, on the one hand, a standardized test problem description system and a measurement system can be established for complex problems on the road; on the other hand, the ability measurement method based on the element dimension can be used to realize the data-driven test case scenario and the expansion of the test problem coverage domain to "unknown problems". However, in the autonomous driving scenario library, due to insufficient historical road test data, the coverage rate of some driving scenarios is low, which is not conducive to comprehensive testing of the algorithm. Therefore, a solution needs to be proposed. Summary of the invention

[0004] Multiple aspects of the present application provide a method, device and storage medium for processing autonomous driving scene data, which are used to generate autonomous driving scene data based on traffic data to increase the coverage of autonomous driving scene data for different autonomous driving scenarios.

[0005] An embodiment of the present application provides a method for processing autonomous driving scene data, including: acquiring static road network data and dynamic traffic data of a road section; identifying at least one traffic trajectory on the road section based on the dynamic traffic data; extracting key frames for the at least one traffic trajectory based on motion features corresponding to the at least one traffic trajectory to obtain at least one group of key frames; wherein the key frames in any traffic trajectory include: data frames corresponding to the changing movements of a moving object; performing decision planning based on the at least one group of key frames and preset constraints to obtain at least one driving trajectory; and determining autonomous driving scene data based on the static road network information and the at least one driving trajectory.

[0006] Optionally, identifying at least one traffic trajectory on the road section based on the dynamic traffic data includes: determining multiple frames of data contained in the dynamic traffic data; identifying the positions of mobile objects in the multiple frames of data to obtain a position sequence of at least one mobile object; fitting the traffic trajectory of each of the at least one mobile object based on the position sequence of each of the at least one mobile object; and optimizing the traffic trajectory of each of the at least one mobile object to obtain the at least one traffic trajectory.

[0007] Optionally, optimizing the traffic trajectory of each of the at least one moving object includes: denoising the traffic trajectory of each of the at least one moving object to obtain a denoised traffic trajectory; determining whether there is a similar trajectory in the denoised traffic trajectory; and if there is a similar trajectory, fusing the similar trajectory.

[0008] Optionally, the traffic trajectory of each of the at least one moving object is denoised to obtain a denoised traffic trajectory, including: for any moving object among the at least one moving object, if there is an interrupted traffic trajectory in the traffic trajectory of the moving object, deleting the interrupted traffic trajectory; and / or, if there is a jittery traffic trajectory in the traffic trajectory of the moving object, smoothing the jittery traffic trajectory; and / or, if there is a reverse traffic trajectory in the traffic trajectory of the moving object, and the reverse traffic trajectory is a pseudo-reverse trajectory, deleting the reverse traffic trajectory.

[0009] Optionally, determining whether there are similar trajectories in the denoised traffic trajectories includes: for any two trajectories in the denoised traffic trajectories, obtaining data features of the any two trajectories from the dynamic traffic data; the data features include: at least one of a lateral distance, a longitudinal distance, a lane number, a vehicle speed, a vehicle acceleration, a vehicle model, and a vehicle geometric size of the any two trajectories; according to the data features of the any two trajectories, obtaining a similarity score between the any two trajectories; if the similarity score is greater than a set similarity threshold and there is a spatiotemporal overlap between the any two trajectories, determining that the any two trajectories are similar trajectories.

[0010] Optionally, based on the motion features corresponding to the at least one traffic trajectory, key frames are extracted from the at least one traffic trajectory to obtain at least one group of key frames, including: for any traffic trajectory of the at least one traffic trajectory, identifying from the dynamic traffic data the behavior of a moving object corresponding to the traffic trajectory; based on the behavior of the moving object, key frames are extracted from the dynamic traffic data corresponding to the traffic trajectory to obtain a group of key frames corresponding to the traffic trajectory.

[0011] Optionally, the behavior includes: at least one of stopping behavior, going against the flow, changing lanes, sudden acceleration behavior and sudden deceleration behavior; according to the behavior of the moving object, key frames are extracted from the dynamic traffic data corresponding to the traffic trajectory to obtain a group of key frames corresponding to the traffic trajectory, including: according to the behavior, at least one key position of the moving object is determined; the at least one key position includes: at least one of a position where the speed change amplitude is greater than a first amplitude threshold, a position where the acceleration change amplitude is greater than a second amplitude threshold, and a position where the direction angle change amplitude is greater than a third amplitude threshold; in the dynamic traffic data corresponding to the traffic trajectory, data frames corresponding to the at least one key position are extracted to obtain a group of key frames corresponding to the traffic trajectory.

[0012] Optionally, in the dynamic traffic data corresponding to the traffic trajectory, extracting a data frame corresponding to the at least one key position, including: for any key position of the at least one key position, obtaining from the dynamic traffic data a relative position relationship between the moving object and the surrounding moving objects when the moving object is at the key position; based on the relative position relationship, determining whether the relative distance between the moving object and the surrounding moving objects is greater than a set distance threshold; if so, extracting a data frame corresponding to the key position in the dynamic traffic data corresponding to the traffic trajectory.

[0013] Optionally, decision planning is performed according to the at least one group of key frames and preset constraints to obtain at least one driving trajectory, including: under the constraints of the constraints, path planning and speed planning are performed according to the at least one group of key frames respectively to obtain the driving trajectory corresponding to the at least one group of key frames; the constraints include at least one of the following: the deviation between the position of the driving trajectory at the key frame and the position perception result corresponding to the key frame is less than a set deviation threshold, the difference between the speed of the driving trajectory at the key frame and the speed perception result corresponding to the key frame is less than a set difference threshold, the position of the driving trajectory is continuous, the speed of the driving trajectory is continuous, and the acceleration of the driving trajectory is continuous.

[0014] Optionally, it also includes: adjusting the constraint conditions; under the constraints of the adjusted constraint conditions, performing path planning and speed planning according to the at least one group of key frames, and using preset disturbance parameters to perturb the path planning results and / or speed planning results to obtain generalized driving trajectories corresponding to each of the at least one group of key frames.

[0015] An embodiment of the present application also provides an electronic device, comprising: a memory and a processor; the memory is used to store one or more computer instructions; the processor is used to execute the one or more computer instructions to: execute the steps in the method provided in the embodiment of the present application.

[0016] The embodiment of the present application also provides a computer-readable storage medium storing a computer program, which can implement the steps in the method provided in the embodiment of the present application when the computer program is executed by a processor.

[0017] In the embodiment of the present application, the autonomous driving scene data is obtained by analyzing and processing the roadside perception data of the road section, and the autonomous driving scene data in a variety of scenarios can be obtained without the need to conduct on-road testing of the autonomous driving vehicle. In addition, by extracting key frames from the traffic trajectory and planning the driving trajectory based on the key frames and constraint conditions, the variability and randomness of the autonomous driving scene data can be increased, thereby deriving driving trajectories in a variety of different scenarios, so as to increase the coverage of the autonomous driving scene data for different autonomous driving scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0019] Figure 1 A flowchart of a method for processing autonomous driving scene data provided by an exemplary embodiment of the present application;

[0020] Figure 2 A schematic diagram of generating and optimizing a traffic path provided for an exemplary embodiment of the present application;

[0021] Figure 3 A flowchart of a method for processing autonomous driving scene data provided by another exemplary embodiment of the present application;

[0022] Figure 4 A schematic diagram of the structure of an electronic device provided as an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0024] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "said" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings, and "multiple" generally includes at least two, but does not exclude the inclusion of at least one.

[0025] It should be understood that the term "and / or" used in this article is only a description of the association relationship of 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.

[0026] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a product or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such a product or system. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the product or system including the elements.

[0027] In the autonomous driving scenario library, due to insufficient historical road test data, the coverage of some driving scenarios is low, which is not conducive to comprehensive testing of the algorithm. In response to the above technical problems, in some embodiments of the present application, a solution is provided, which obtains autonomous driving scenario data by analyzing and mining the roadside perception data of the road section, thereby obtaining autonomous driving scenario data in rich scenarios without the need to conduct on-road testing of the autonomous driving algorithm.

[0028] The technical solutions provided by various embodiments of the present application are described in detail below in conjunction with the accompanying drawings.

[0029] Figure 1 is a flow chart of a method for processing autonomous driving scene data provided by an exemplary embodiment of the present application. The method may include: Figure 1 Steps shown:

[0030] Step 101: Obtain static road network data and dynamic traffic data of a road section.

[0031] Step 102: Identify at least one traffic trajectory on the road segment based on the dynamic traffic data.

[0032] Step 103: extract key frames from the at least one traffic trajectory according to the motion features corresponding to the at least one traffic trajectory to obtain at least one group of key frames; wherein the key frames in any traffic trajectory include: data frames corresponding to the changing motion of the moving object.

[0033] Step 104: Decision planning is performed according to the at least one set of key frames and preset constraints to obtain at least one driving trajectory.

[0034] Step 105: Determine the automatic driving scenario data based on the static road network information and the at least one driving trajectory.

[0035] In this embodiment, the static road network data of the road section can be obtained from the high-precision map data and / or roadside perception data corresponding to the road section. Among them, the roadside perception data refers to the data obtained by collecting multi-dimensional information on the road section through the perception equipment (i.e., roadside perception equipment) deployed around the road section (for example, on both sides of the road section or above the road section). Among them, the perception equipment may include but is not limited to millimeter wave radar, laser radar, high-definition camera, etc., and this embodiment does not limit this.

[0036] Among them, static road network data refers to data used to describe the road network characteristics of a road section, including but not limited to at least one of the lane type, lane attributes, lane lines, intersections, guardrails and traffic lights of the road section.

[0037] The dynamic traffic data may be obtained from the roadside sensing data of the road section. The dynamic traffic data refers to the data generated by the traffic behavior of the moving objects on the road section, including but not limited to at least one of the type, movement position, movement speed and appearance characteristics of the moving objects on the road section. The moving objects on the road section include but are not limited to at least one of motor vehicles, non-motor vehicles and pedestrians.

[0038] According to the dynamic traffic data, at least one traffic trajectory on the road section can be identified. The traffic trajectory refers to the curve formed by the change of the spatial position of the moving object over time when the moving object performs traffic behavior. When identifying the traffic trajectory of any moving object according to the dynamic traffic data, the discretely distributed motion position of the moving object can be obtained from the dynamic traffic data, and the discretely distributed motion position can be curve fitted in chronological order to obtain the traffic trajectory. In the process of curve fitting, the discretely distributed motion position can be corrected or optimized to make the fitted traffic trajectory more consistent with the real traffic scene. This section will be specifically introduced in the subsequent embodiments and will not be repeated here.

[0039] After determining at least one traffic trajectory, key frames can be extracted from the dynamic traffic data according to the motion features corresponding to the at least one traffic trajectory. The motion features corresponding to any traffic trajectory refer to the features of the moving object in terms of motion action, motion speed, etc. as reflected by the traffic trajectory. The key frames include data frames corresponding to the changing actions of the moving object in the traffic trajectory. For example, the key frame can be a data frame in the traffic trajectory corresponding to the lane change action of the moving object, or it can be a data frame in the traffic trajectory corresponding to the U-turn action of the moving object. It should be understood that when the key frame corresponds to the changing action of the moving object, the change amount of the trajectory between two adjacent key frames is small. Therefore, the amount of stored data can be reduced and the storage cost can be optimized by storing key frames instead of directly storing traffic trajectories.

[0040] Among them, the key frames extracted from the dynamic traffic data can be used for path planning to obtain a path that can be used for testing the autonomous driving algorithm, hereinafter referred to as a driving trajectory. In this embodiment, decision planning can be performed based on the at least one set of key frames and preset constraints to obtain at least one driving trajectory. Among them, decision planning refers to planning a reasonable path for the passage of moving objects based on input information. In this embodiment, the input information of any decision planning includes at least a set of key frames and preset constraints, which are used to fit the driving path based on a set of key frames and make the driving path meet the constraints. Among them, the constraints may include but are not limited to at least one of speed conditions and position conditions. Among them, a set of key frames can decide to plan different driving trajectories under different constraints. Furthermore, by changing the constraints, paths under a variety of different scenarios can be generated to enrich the autonomous driving scene data.

[0041] After deciding and planning at least one driving trajectory, the autonomous driving scene data can be determined based on the static road network information and the at least one driving trajectory. Among them, the static road network can be converted into a road network file in a specified format required by the autonomous driving simulation platform, and the dynamic elements such as the driving trajectory can be converted into a dynamic scene file in a specified format required by the autonomous driving simulation platform. The road network data and the dynamic scene file constitute the autonomous driving scene data. The autonomous driving scene data can be used as a test case, simulation case, and training case for the algorithm model of autonomous driving to improve the adaptability of the autonomous driving vehicle in different scenarios. For example, when the autonomous driving algorithm is updated, the test case can be used on the simulation platform to test the algorithm, and evaluate the safety, comfort and other related indicators of the main vehicle in the scenario, so as to evaluate the autonomous driving algorithm.

[0042] In this embodiment, the autonomous driving scene data is obtained by analyzing and processing the roadside perception data of the road section, and the autonomous driving scene data in a variety of scenarios can be obtained without the need to conduct on-road testing of the autonomous driving algorithm. In addition, by extracting key frames from traffic trajectories and planning driving trajectories based on key frames and constraint conditions, the variability and randomness of the autonomous driving scene data can be increased, thereby deriving driving trajectories in a variety of different scenarios, so as to increase the coverage of the autonomous driving scene data for different autonomous driving scenarios.

[0043] In some embodiments, the roadside perception device reports the roadside perception data collected within a set time period to the server under the triggering of a set event. The electronic device used to generate autonomous driving scene data can obtain roadside perception data from the server and execute the methods provided in the embodiments of the present application to generate autonomous driving scene data. Among them, the set event can be an event of receiving a user instruction, an event extracted from the collected data, or an event of sensing a specific road condition. For example, in some embodiments, the roadside perception device can perform event detection based on the perceived data during the process of sensing the dynamic traffic data of the road section. If the roadside perception device detects an abnormal event on the road section, the roadside perception data collected within the specified time and space range before and after the abnormal event can be extracted. Among them, abnormal events can include but are not limited to: traffic accidents, severe weather events, maintenance and construction events, and emergency events. Among them, the specified time range can be set according to needs. For example, the roadside perception device can extract roadside perception data from 5 seconds before the event to 10 seconds after the event.

[0044] Based on this implementation method, the dynamic traffic data when abnormal events occur can be analyzed to obtain the driving trajectory when the abnormal events occur, so that the driving trajectory can cover the traffic scenes where the abnormal events occur, thereby improving the ability of the autonomous driving algorithm to respond to the traffic scenes corresponding to the abnormal events.

[0045] In some exemplary embodiments, at least one traffic track on the road section may be identified based on the dynamic traffic data based on a single-frame positioning and multi-frame tracking method. Figure 2 An exemplary description is given.

[0046] Optionally, multiple frames of data contained in the dynamic traffic data may be determined. The acquisition time of each frame of data is different. The position of the mobile object in the multiple frames of data is identified to obtain a position sequence of at least one mobile object. In this step, the mobile object in each frame of data may be first detected and the position of the mobile object may be identified. The detection operation of the mobile object may be implemented based on the target detection algorithm, which is not described here. When identifying the position of the mobile object, the position of the mobile object relative to the roadside sensing device may be calculated, and the position of the mobile object relative to the roadside sensing device may be mapped according to the position of the roadside sensing device in the world coordinate system to obtain the position of the mobile object in the world coordinate system. For example, when the roadside sensing device is a high-definition camera, multiple continuous video frames may be obtained from a video of a certain length captured by the high-definition camera. The mobile objects in the multiple video frames are identified respectively, and the position of the mobile object in the image coordinate system is detected. According to the calibration parameters of the camera, the position of the mobile object in the image coordinate system may be converted into the position of the mobile object in the camera coordinate system. According to the position of the high-definition camera in the world coordinate system, the position of the moving object in the camera coordinate system can be converted into the position of the moving object in the world coordinate system, which will not be described in detail.

[0047] By identifying the position of a moving object in a continuous multi-frame data, a position sequence of the moving object can be obtained. The position sequence is arranged in chronological order, such as Figure 2 According to the position sequence of the at least one moving object, the traffic track of the at least one moving object can be fitted.

[0048] In some special events, accidents or complex working conditions, due to the limitations of roadside sensing equipment, roadside sensing data may have some missed detections, false detections, and other abnormal collection situations, and the fitted traffic trajectory may have ID (Identity document) switching, trajectory jitter or trajectory abnormalities. Figure 2 As shown in FIG. 1 , the traffic trajectory obtained by fitting the discrete position sequence has some trajectory jitters.

[0049] In view of the above situation, the traffic trajectory of each of the at least one moving object may be optimized to obtain the at least one traffic trajectory.

[0050] Optionally, optimizing the traffic trajectory of the at least one moving object may include: performing denoising processing on the traffic trajectory of the at least one moving object to obtain a denoised traffic trajectory; Figure 2 The smooth trajectory after trajectory optimization is shown. The denoising process may include at least one of the following implementations:

[0051] Optionally, for any of the at least one moving object, if there is an interrupted traffic trajectory in the traffic trajectory of the moving object, the interrupted traffic trajectory is deleted. The interrupted traffic trajectory is usually manifested as: the starting point of the trajectory suddenly appears in the middle of the road section, or the end point of the trajectory suddenly appears in the middle of the road section. Therefore, it can be determined whether the traffic trajectory is an interrupted traffic trajectory by identifying whether the starting point of any traffic trajectory is located at a non-entrance position of the road section, and / or identifying whether the end point of the road section is located at a non-exit position of the road section.

[0052] Optionally, for any of the at least one moving object, if there is a jittery traffic trajectory in the traffic trajectory of the moving object, the jittery traffic trajectory is smoothed. The jittery traffic trajectory means that some positions in the traffic trajectory have obvious oscillations or fluctuations. During the optimization process, the positions with oscillations or fluctuations can be smoothed to make the trajectory more consistent with the actual traffic behavior.

[0053] Optionally, for any one of the at least one moving objects, if there is a reverse traffic trajectory in the traffic trajectory of the moving object, and the reverse traffic trajectory is a pseudo-reverse trajectory, the reverse traffic trajectory is deleted. Wherein, the reverse traffic trajectory is usually manifested as: the forward direction of the traffic trajectory is inconsistent with the driving direction of the road section. The reverse traffic trajectory includes: a true reverse trajectory and a pseudo-reverse trajectory. Wherein, the conventional abnormal trajectory means that the moving object has actually performed a reverse behavior. In the case of actual reverse behavior, the probability of a conflict between the traffic trajectory of the moving object and the traffic trajectory of other moving objects is small, so the true reverse trajectory can be retained to generate the automatic driving scene data corresponding to the reverse scene. The pseudo-reverse trajectory means that due to the limited detection of the roadside perception device, the non-reverse behavior of the moving object is identified as a reverse behavior, but the moving object has not actually performed a reverse behavior. In the case of no actual reverse behavior, the probability of a conflict between the traffic trajectory of the moving object and the traffic trajectory of other moving objects is large, so the pseudo-reverse trajectory can be deleted.

[0054] In some embodiments, after obtaining the denoised traffic trajectory, it can be determined whether there are similar trajectories in the denoised traffic trajectory. Optionally, when determining whether there are similar trajectories in the denoised traffic trajectory, it can be determined whether any two trajectories are similar. For any two trajectories in the denoised traffic trajectory, the data features of the any two trajectories are obtained from the dynamic traffic data; the data features include: at least one of the lateral distance, longitudinal distance, lane number, vehicle speed, vehicle acceleration, vehicle model and vehicle geometric dimensions of the any two trajectories. According to the data features of the any two trajectories, the similarity score between the any two trajectories is obtained. Wherein, when multiple data features are included, the similarity scores of the any two trajectories on the multiple data features can be calculated, and the weighted sum is performed according to the similarity scores on the multiple data features to obtain the similarity scores of the any two trajectories. Wherein, the weighted weights of the multiple data features can be the same or different, and can be set specifically according to the calculation requirements. If the similarity score of the any two trajectories is greater than the set similarity threshold, the any two trajectories are determined to be similar trajectories. Based on this implementation, trajectories with temporal and spatial overlaps collected by the same or different roadside sensing devices can be identified. After identifying similar trajectories, it can be determined whether the similar trajectories have temporal and spatial overlaps. Among them, the temporal and spatial overlaps between any two trajectories means that any two trajectories overlap in time and in space. When any two similar trajectories overlap in time and space, it can be considered that the any two trajectories are traffic trajectories formed by the same moving object sensed by different road test sensing devices. If any two traffic trajectories are similar trajectories, and the any two trajectories overlap in time and space, the any two similar trajectories can be fused to reduce the probability of the ID switching problem.

[0055] The above embodiments may be executed individually or in combination, depending on the actual noise conditions of the traffic trajectory, and this embodiment does not impose any limitation thereto.

[0056] In some exemplary embodiments, according to the motion characteristics corresponding to the at least one traffic trajectory, key frames are extracted from the dynamic traffic data to obtain at least one set of key frames, and the behavior characteristics of the moving object can be identified, and key frames can be extracted according to the behavior characteristics. The following will take any one of the at least one traffic trajectory as an example for exemplary description.

[0057] For any traffic trajectory of the at least one traffic trajectory, the behavior of the moving object corresponding to the traffic trajectory can be identified from the dynamic traffic data; according to the behavior of the moving object, key frames can be extracted from the traffic trajectory to obtain a set of key frames corresponding to the traffic trajectory.

[0058] In this embodiment, the behavior of the moving object refers to a behavior of the moving object that changes relative to the behavior in the previous historical time period. Optionally, the behavior includes at least one of: stopping behavior, reverse behavior, lane changing behavior, rapid acceleration behavior, and rapid deceleration behavior.

[0059] Accordingly, when a key frame is extracted from the dynamic traffic data corresponding to the traffic trajectory according to the behavior of the mobile object and a group of key frames corresponding to the traffic trajectory are obtained, at least one key position of the mobile object can be determined according to the behavior. The at least one key position includes at least one of: a position where the speed change amplitude is greater than the first amplitude threshold, a position where the acceleration change amplitude is greater than the second amplitude threshold, and a position where the direction angle change amplitude is greater than the third amplitude threshold. For example, when the mobile object is detected to have stopped, the position where the speed change amplitude of the mobile object is greater than the first amplitude threshold can be determined according to the position where the stopping behavior occurs. For example, when the mobile object is detected to have reverse behavior, the position where the acceleration change amplitude of the mobile object is greater than the second amplitude threshold can be determined according to the position where the reverse behavior occurs. For another example, when the mobile object is detected to have changed lanes, the position where the direction angle change amplitude of the mobile object is greater than the third amplitude threshold can be determined according to the position where the lane changing behavior occurs. After determining the position corresponding to the at least one key position, the data frame corresponding to the at least one key position can be extracted from the dynamic traffic data corresponding to the traffic trajectory to obtain a group of key frames corresponding to the traffic trajectory.

[0060] It should be noted that when a moving object is at a key position, if the surrounding moving objects are close to it, a key frame is taken here, and there may be a situation where the path passing through the key frame intersects with the trajectory of the surrounding moving objects. Based on this, when selecting a key frame according to the position point corresponding to the key position, the surrounding environment data of the moving object at the key position can be further obtained. Based on the surrounding environment data, the relative position relationship between the moving object and the surrounding moving objects can be determined, and based on the relative position relationship, it can be determined whether the position point corresponding to the key position on the traffic trajectory can be used as a key frame.

[0061] Optionally, for any key position among the at least one key position, the relative position relationship between the moving object and the surrounding moving objects when the moving object is at the key position can be obtained from the dynamic traffic data; based on the relative position relationship, it is determined whether the relative distance between the moving object and the surrounding moving objects is greater than a set distance threshold; if so, the data frame corresponding to the key position is extracted as a key frame from the dynamic traffic data corresponding to the traffic trajectory; otherwise, it is not necessary to extract the data frame corresponding to the key position from the dynamic traffic data corresponding to the traffic trajectory. Based on this implementation, the position points that are close to the moving object and the surrounding moving objects can be bypassed in the process of extracting the key frame, thereby reducing the risk of errors in the determined path and improving the availability of the determined path.

[0062] After obtaining at least one set of key frames based on the methods provided in the aforementioned embodiments, decision planning can be performed based on the at least one set of key frames to obtain a driving trajectory.

[0063] In some exemplary embodiments, when decision planning is performed according to the at least one group of key frames and preset constraints, path planning and speed planning can be performed according to the at least one group of key frames respectively under the constraints of the constraints, and the driving trajectory corresponding to each of the at least one group of key frames can be obtained. Wherein, the constraints include at least one of the following: the deviation between the position of the driving trajectory at the key frame and the position perception result corresponding to the key frame is less than the set deviation threshold, the difference between the speed of the driving trajectory at the key frame and the speed perception result corresponding to the key frame is less than the set difference threshold, the position of the driving trajectory is continuous, the speed of the driving trajectory is continuous, and the acceleration of the driving trajectory is continuous. Based on the constraints, the deviation between the path decided by the planning and the path actually perceived by the roadside perception device can be within a certain range, and the generated driving trajectory can be smooth and continuous, thereby improving the quality of the driving trajectory.

[0064] It is worth noting that, in some embodiments, different driving trajectories can be generated by adjusting the constraint conditions. An exemplary description will be given below.

[0065] Optionally, adjusting the constraint condition may include: adjusting the combination of the above-mentioned multiple constraint conditions. For example, the constraint condition may be adjusted to: the deviation between the position of the driving trajectory at the key frame and the position perception result corresponding to the key frame is less than a set deviation threshold, and the position of the driving trajectory is continuous. For another example, the constraint condition may be adjusted to: the difference between the speed of the driving trajectory at the key frame and the speed perception result corresponding to the key frame is less than a set difference threshold, and the speed of the driving trajectory is continuous.

[0066] Optionally, adjusting the constraint conditions may include: adjusting the deviation threshold and / or difference threshold recorded above. For example, the constraint conditions may be adjusted to: the deviation between the position of the driving trajectory at the key frame and the position perception result corresponding to the key frame is less than the increased deviation threshold, and the difference between the speed of the driving trajectory at the key frame and the speed perception result corresponding to the key frame is less than the increased difference threshold. Furthermore, a driving trajectory with higher randomness may be planned based on the relaxed constraint conditions.

[0067] Under the constraints of the adjusted constraints, path planning and speed planning can be performed according to the at least one group of key frames, and the path planning results and / or speed planning results can be disturbed using preset disturbance parameters to obtain generalized driving trajectories corresponding to the at least one group of key frames.

[0068] Among them, the disturbance parameters corresponding to the path planning result may include: at least one of the direction disturbance parameters and the path detour disturbance parameters. For example, the path planning result may be disturbed by changing lanes or the path planning result may be disturbed by detours (for example, a detour path is generated between two key frames) to simulate a driving path that changes lanes or detours midway. Among them, the disturbance parameters corresponding to the speed planning result may include: at least one of the acceleration disturbance parameters and the deceleration disturbance parameters. For example, the speed planning result may be disturbed by acceleration and deceleration to simulate a driving trajectory of sudden acceleration or sudden braking. In this embodiment, the constraints are adjusted and the disturbance factors are increased during the decision-making planning process, so that the generated driving trajectory can be within a feasible solution range without reaching the solution. On the one hand, the number of iterations can be reduced, and on the other hand, the scenes under different parameter distributions can be simulated. For example, in some scenarios, when testing the algorithm of the autonomous driving vehicle (main vehicle), the effect of the vehicles around the main vehicle driving according to the generalized driving trajectory generated by the above-mentioned embodiment can be simulated, so that the trajectory of the vehicles around the main vehicle can be flexibly changed within a controllable range, thereby increasing the test simulation of different scenarios. For example, an acceleration driving trajectory can be generated according to the acceleration disturbance parameters, and the effect of vehicles around the host vehicle running along the acceleration driving trajectory can be simulated to test the performance of the host vehicle's algorithm when the surrounding vehicles are accelerating.

[0069] In this implementation, by adjusting the constraints and adding disturbance parameters to the decision-making planning results, richer driving scenarios are generalized, thereby mining autonomous driving scenario data in more scenarios based on the existing roadside perception data, and increasing the coverage of autonomous driving scenario data for a variety of different autonomous driving scenarios, thereby improving the autonomous driving algorithm's ability to respond to different scenarios.

[0070] In some optional embodiments, after the driving trajectory is obtained based on the aforementioned embodiments, a rationality judgment can be made for the driving trajectory. Taking the mobile object as a vehicle as an example, the rationality judgment can be implemented based on the vehicle dynamics model. When making a rationality judgment on any driving trajectory, it can be judged whether the speed change, acceleration change, and angular velocity change of the driving trajectory conform to the constraints of the vehicle dynamics model. If the driving trajectory conforms to the constraints of the vehicle dynamics model, and the loss function between the driving trajectory and the trajectory actually perceived by the roadside sensing device converges to a certain range, the driving trajectory is regarded as a reasonable driving trajectory.

[0071] After screening out reasonable driving trajectories based on the aforementioned embodiments, the driving trajectories and static road network information can be processed into autonomous driving scenarios for use in the simulation of autonomous driving algorithms, which will not be described in detail.

[0072] The following will be combined Figure 3 , the method for obtaining autonomous driving scene data provided in an embodiment of the present application is further exemplified.

[0073] like Figure 3 As shown, the roadside perception device can extract abnormal events in the process of perceiving the road section data, and report the roadside perception data corresponding to the abnormal events. Among them, abnormal events include: parking, reverse driving, collision, construction and other events. In the process of generating autonomous driving scene data, the roadside perception data can be extracted from the high-precision road network to obtain static road network data. Among them, if the roadside perception data has collected the static information of the high-precision road network, the road section element information can be extracted from the static information, including lane type, lane attribute, lane line, intersection, guardrail, traffic light and other road section elements; if the roadside perception data has not directly collected the static information of the high-precision road network, the above-mentioned road section elements can be extracted from the video clips in the roadside perception data. In the video clips, the above-mentioned road section element information can be extracted based on the image processing algorithm, and after the road section element information is extracted, it can be adjusted according to the actual situation of the road section.

[0074] In the process of generating autonomous driving scene data, dynamic elements can be extracted from roadside perception data. For holographic road sections (i.e., sections where multiple roadside perception devices such as millimeter-wave radars, lidars, and cameras are deployed), road test perception devices can be used to detect, track, and locate moving objects, thereby obtaining relevant dynamic elements. Among them, moving objects include motor vehicles, non-motor vehicles, and pedestrians; for non-holographic road sections (i.e., sections where radar equipment is not deployed), video clips collected from cameras on non-holographic road sections can be obtained, and the static road network data extracted in the previous steps can be mapped to the images in the video clips, and moving objects in the video clips can be detected, tracked, located, and coordinate converted.

[0075] After extracting dynamic elements, dynamic trajectories can be generated based on the dynamic elements. In events, accidents, and complex working conditions, due to the limitations of cameras and radars, roadside perception results often have false detections, ID switching, trajectory jitter, trajectory anomalies, etc., which cannot meet the usage standards of the scene library. Therefore, after generating the dynamic trajectory, dynamic trajectory optimization operations can be performed. Figure 3 As shown, dynamic trajectory optimization includes: denoising, fusion, sampling and completion operations.

[0076] In the process of trajectory denoising, the rationality of the trajectory and whether the trajectory conflicts with other reasonable trajectories can be judged to determine whether there is trajectory noise on the trajectory. If there is, the trajectory noise can be removed. The trajectory noise includes incomplete or abnormal trajectories such as sudden appearance or disappearance, jitter, stop, and reverse.

[0077] In the process of trajectory fusion, multi-frame data features (including lateral distance, longitudinal distance, lane number, speed, acceleration, vehicle type, length, width and height, etc.) can be used to judge the trajectory similarity and solve the optimal result to obtain similar trajectories. After determining similar trajectories, similar trajectories with temporal and spatial overlap captured by the same / different cameras can be fused to ensure the global uniqueness of the vehicle ID and reduce the probability of vehicle ID switching.

[0078] In the process of trajectory sampling, keyframes can be extracted from the dynamic element data corresponding to the dynamic trajectory. In the process of keyframe extraction, the vehicle's stopping, reverse driving, lane changing and other behaviors can be analyzed, and keyframes can be extracted based on information such as vehicle speed, acceleration, and direction angle changes. In the process of keyframe extraction, the vehicle's key position information can be determined, and keyframes can be selected based on the relative position relationship between the vehicle and surrounding vehicles.

[0079] Among them, in the process of completion, decision planning can be carried out according to the sampled key frames. In the process of decision planning, the trajectory optimization algorithm can be used to fit the corresponding position points of the key frames, and the position and speed can be planned at the same time. In the process of position and speed planning, constraints can be set to ensure that the deviation between the fitting result and the perceived real trajectory is within the set range, thereby ensuring the quality of trajectory output. Among them, the constraints may include: the trajectory passes through the key frame, the deviation between the speed and position of the trajectory and the perception result in the key frame is within a certain range, and the trajectory position is smooth and the speed is continuous. Based on trajectory fusion, key frame extraction and decision planning, it is possible to better fuse the perception results of multiple overlapping areas across point trajectories, and further smooth and repair the trajectory to obtain a high-quality driving trajectory.

[0080] After planning the trajectory based on the above steps, the scenario library can be constructed. In this step, the extracted high-precision road network section element information and the fitted driving trajectory can be converted into high-precision road networks and scenarios in corresponding formats according to the requirements of each autonomous driving simulation platform, and a scenario library can be formed through long-term processing and accumulation.

[0081] Based on this scenario library, autonomous driving simulation can be performed. During the simulation process, the scenario library can be used to simulate various events and accident scenarios through the autonomous driving platform, thereby realizing the evaluation of perception, decision-making, and control algorithms.

[0082] It should be noted that the execution subject of each step of the method provided in the above embodiment can be the same device, or the method can be executed by different devices. For example, the execution subject of steps 101 to 104 can be device A; for another example, the execution subject of steps 101 and 102 can be device A, and the execution subject of step 103 can be device B; and so on.

[0083] In addition, in some of the processes described in the above embodiments and the accompanying drawings, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel, and the sequence numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the sequence numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to be different types.

[0084] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0085] Figure 4 A schematic diagram of the structure of an electronic device provided by an exemplary embodiment of the present application is shown. Figure 4 As shown, the electronic device includes: a memory 401 , a processor 402 and a communication component 403 .

[0086] The memory 401 is used to store computer programs and can be configured to store various other data to support operations on the electronic device. Examples of such data include instructions for any application program or method operating on the electronic device.

[0087] The processor 402 is coupled to the memory 401 and is used to execute the computer program in the memory 401, so as to: obtain static road network data and dynamic traffic data of a road section; identify at least one traffic trajectory on the road section according to the dynamic traffic data; extract key frames for the at least one traffic trajectory according to the motion characteristics corresponding to the at least one traffic trajectory to obtain at least one group of key frames; wherein the key frames in any traffic trajectory include: data frames corresponding to the changing movements of the moving object; perform decision planning according to the at least one group of key frames and preset constraints to obtain at least one driving trajectory; determine the automatic driving scene data according to the static road network information and the at least one driving trajectory.

[0088] Optionally, when the processor 402 identifies at least one traffic trajectory on the road section based on the dynamic traffic data, it is specifically used to: determine multiple frames of data contained in the dynamic traffic data; identify the positions of mobile objects in the multiple frames of data to obtain a position sequence of at least one mobile object; fit the traffic trajectory of each of the at least one mobile object based on the position sequence of each of the at least one mobile object; and optimize the traffic trajectory of each of the at least one mobile object to obtain the at least one traffic trajectory.

[0089] Optionally, when optimizing the traffic trajectory of each of the at least one moving object, the processor 402 is specifically used to: denoise the traffic trajectory of each of the at least one moving object to obtain a denoised traffic trajectory; determine whether there is a similar trajectory in the denoised traffic trajectory; if there is a similar trajectory, fuse the similar trajectory.

[0090] Optionally, when the processor 402 performs denoising processing on the traffic trajectory of each of the at least one mobile object to obtain the denoised traffic trajectory, it is specifically used to: for any mobile object among the at least one mobile object, if there is an interrupted traffic trajectory in the traffic trajectory of the mobile object, delete the interrupted traffic trajectory; and / or, if there is a jittery traffic trajectory in the traffic trajectory of the mobile object, smooth the jittery traffic trajectory; and / or, if there is a reverse traffic trajectory in the traffic trajectory of the mobile object, and the reverse traffic trajectory is a pseudo-reverse trajectory, delete the reverse traffic trajectory.

[0091] Optionally, when determining whether there are similar trajectories in the denoised traffic trajectory, the processor 402 is specifically used to: for any two trajectories in the denoised traffic trajectory, obtain data features of the any two trajectories from the dynamic traffic data; the data features include: at least one of the lateral distance, longitudinal distance, lane number, vehicle speed, vehicle acceleration, vehicle model and vehicle geometric dimensions of the any two trajectories; based on the data features of the any two trajectories, obtain a similarity score between the any two trajectories; if the similarity score is greater than a set similarity threshold and there is a spatiotemporal overlap between the any two trajectories, determine that the any two trajectories are similar trajectories.

[0092] Optionally, the processor 402 extracts key frames for the at least one traffic trajectory according to the motion features corresponding to the at least one traffic trajectory to obtain at least one group of key frames, including: for any traffic trajectory of the at least one traffic trajectory, identifying the behavior of the moving object corresponding to the traffic trajectory from the dynamic traffic data; and extracting key frames from the dynamic traffic data corresponding to the traffic trajectory according to the behavior of the moving object to obtain a group of key frames corresponding to the traffic trajectory.

[0093] Optionally, the behavior includes: at least one of stopping behavior, going against the flow, changing lanes, sudden acceleration behavior and sudden deceleration behavior; when the processor 402 extracts key frames from the dynamic traffic data corresponding to the traffic trajectory according to the behavior of the moving object to obtain a group of key frames corresponding to the traffic trajectory, it is specifically used to: determine at least one key position of the moving object according to the behavior; the at least one key position includes: at least one of a position where the speed change amplitude is greater than a first amplitude threshold, a position where the acceleration change amplitude is greater than a second amplitude threshold, and a position where the direction angle change amplitude is greater than a third amplitude threshold; in the dynamic traffic data corresponding to the traffic trajectory, extract the data frame corresponding to the at least one key position to obtain a group of key frames corresponding to the traffic trajectory.

[0094] Optionally, when the processor 402 extracts the data frame corresponding to the at least one key position in the dynamic traffic data corresponding to the traffic trajectory, it is specifically used to: for any key position among the at least one key position, obtain from the dynamic traffic data the relative position relationship between the moving object and the surrounding moving objects when the moving object is at the key position; based on the relative position relationship, determine whether the relative distance between the moving object and the surrounding moving objects is greater than a set distance threshold; if yes, extract the data frame corresponding to the key position in the dynamic traffic data corresponding to the traffic trajectory.

[0095] Optionally, when the processor 402 performs decision planning according to the at least one group of key frames and preset constraints to obtain at least one driving trajectory, it is specifically used to: under the constraints of the constraints, perform path planning and speed planning according to the at least one group of key frames respectively to obtain the driving trajectory corresponding to each of the at least one group of key frames; the constraints include at least one of the following: the deviation between the position of the driving trajectory at the key frame and the position perception result corresponding to the key frame is less than a set deviation threshold, the difference between the speed of the driving trajectory at the key frame and the speed perception result corresponding to the key frame is less than a set difference threshold, the position of the driving trajectory is continuous, the speed of the driving trajectory is continuous, and the acceleration of the driving trajectory is continuous.

[0096] Optionally, the processor 402 is also used to: adjust the constraint conditions; under the constraints of the adjusted constraint conditions, perform path planning and speed planning according to the at least one group of key frames, and use preset disturbance parameters to perturb the path planning results and / or speed planning results to obtain generalized driving trajectories corresponding to each of the at least one group of key frames.

[0097] Further, if Figure 4 As shown, the electronic device also includes: a power supply component 404, a display component 405, an audio component 406 and other components. Figure 4 Only some components are shown schematically, which does not mean that the electronic device only includes Figure 4 Components shown.

[0098] Among them, the memory 401 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0099] The communication component 403 is configured to facilitate wired or wireless communication between the device where the communication component is located and other devices. The device where the communication component is located can access a wireless network based on a communication standard, such as Wi-Fi (wireless network communication technology), 2G (such as Global System for Mobile Communications (GSM)), 3G (such as Wideband Code Division Multiple Access (WCDMA), 4G (such as Long Term Evolution (LTE)), 4G+ (such as upgraded Long Term Evolution (LTE-Advanced, LTE-A)) or 5G (Fifth Generation Mobile Communication Technology), or a combination thereof. In an exemplary embodiment, the communication component receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component can be implemented based on Near Field Communication (NFC) technology, Radio Frequency Identification (RFID) technology, Infrared Data Association (IrDA) technology, Ultra Wide Band (UWB) technology, Bluetooth (BT) technology and other technologies.

[0100] The power supply component 404 is used to provide power to various components of the device where the power supply component is located. The power supply component may include a power management system, one or more power supplies, and other components associated with generating, managing and distributing power to the device where the power supply component is located.

[0101] The display component 405 includes a screen, which may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from a user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of a touch or slide action, but also detect the duration and pressure associated with the touch or slide operation.

[0102] The audio component 406 may be configured to output and / or input audio signals. For example, the audio component includes a microphone (MIC), and when the device where the audio component is located is in an operating mode, such as a call mode, a recording mode, and a speech recognition mode, the microphone is configured to receive an external audio signal. The received audio signal may be further stored in a memory or sent via a communication component. In some embodiments, the audio component also includes a speaker for outputting an audio signal.

[0103] In this embodiment, the autonomous driving scene data is obtained by analyzing and processing the roadside perception data of the road section, and the autonomous driving scene data in a variety of scenarios can be obtained without the need to conduct on-road testing of the autonomous driving vehicle. In addition, by extracting key frames from the traffic trajectory and planning the driving trajectory based on the key frames and constraint conditions, the variability and randomness of the autonomous driving scene data can be increased, thereby deriving driving trajectories in a variety of different scenarios, so as to increase the coverage of the autonomous driving scene data for different autonomous driving scenarios.

[0104] Accordingly, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed, can implement each step that can be executed by an electronic device in the above method embodiment.

[0105] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM (Compact Disc Read-Only Memory), optical storage, etc.) containing computer-usable program codes.

[0106] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0107] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0108] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0109] In a typical configuration, a computing device includes one or more processors (Central Processing Unit, CPU), input / output interface, network interface and memory.

[0110] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0111] Computer-readable media include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, Parallel Random Access Machine (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, CD-ROM, Digital Video Disc (DVD) or other optical storage, magnetic cassettes, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include temporary computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0112] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0113] The above is only the embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for processing autonomous driving scene data, characterized in that: include: Obtain static road network data and dynamic traffic data of the road section; identifying at least one traffic trajectory on the road segment based on the dynamic traffic data; According to the motion features corresponding to the at least one traffic trajectory, key frames are extracted from the at least one traffic trajectory to obtain at least one group of key frames; wherein the key frames in any traffic trajectory include: data frames corresponding to the changing motion of the moving object: Perform decision planning according to the at least one set of key frames and preset constraints to obtain at least one driving trajectory; Determine the automatic driving scenario data based on the static road network information and the at least one driving trajectory.

2. The method according to claim 1, characterized in that Identifying at least one traffic track on the road segment according to the dynamic traffic data, including: Determine a plurality of frames of data included in the dynamic traffic data; Identifying positions of moving objects in the multiple frames of data to obtain a position sequence of at least one moving object; According to the respective position sequences of the at least one moving object, fittingly obtaining the respective traffic trajectories of the at least one moving object; The traffic trajectory of each of the at least one moving object is optimized to obtain the at least one traffic trajectory.

3. The method according to claim 2, characterized in that Optimizing the traffic trajectory of each of the at least one moving object, including: De-noising the traffic trajectory of the at least one moving object to obtain a de-noised traffic trajectory; Determining whether there is a similar trajectory in the denoised traffic trajectory; If similar trajectories exist, the similar trajectories are fused.

4. The method according to claim 3, characterized in that De-noising the traffic trajectory of the at least one moving object to obtain a de-noised traffic trajectory, including: For any one of the at least one moving object, if there is an interrupted traffic track in the traffic track of the moving object, deleting the interrupted traffic track; and / or, If there is a jittery traffic trajectory in the traffic trajectory of the moving object, smoothing the jittery traffic trajectory; and / or, If there is a reverse traffic trajectory in the traffic trajectory of the moving object, and the reverse traffic trajectory is a pseudo reverse traffic trajectory, the reverse traffic trajectory is deleted.

5. The method according to claim 3, characterized in that: Determining whether there is a similar trajectory in the denoised traffic trajectory includes: For any two trajectories in the denoised traffic trajectories, data features of the any two trajectories are obtained from the dynamic traffic data; the data features include: at least one of a lateral distance, a longitudinal distance, a lane number, a vehicle speed, a vehicle acceleration, a vehicle model, and a vehicle geometric size of the any two trajectories; Obtaining a similarity score between the two trajectories according to the data features of the two trajectories; If the similarity score is greater than a set similarity threshold and there is a temporal and spatial overlap between the two arbitrary trajectories, the two arbitrary trajectories are determined to be similar trajectories.

6. The method according to any one of claims 1 to 5, characterized in that: According to the motion feature corresponding to the at least one traffic trajectory, key frames are extracted from the at least one traffic trajectory to obtain at least one group of key frames, including: For any one of the at least one traffic trajectory, identifying a behavior of a moving object corresponding to the traffic trajectory from the dynamic traffic data; According to the behavior of the moving object, key frames are extracted from the dynamic traffic data corresponding to the traffic trajectory to obtain a group of key frames corresponding to the traffic trajectory.

7. The method according to claim 6, characterized in that The behavior includes: at least one of: stopping behavior, reverse behavior, lane changing behavior, sudden acceleration behavior and sudden deceleration behavior; According to the behavior of the moving object, key frames are extracted from the dynamic traffic data corresponding to the traffic trajectory to obtain a set of key frames corresponding to the traffic trajectory, including: Determine at least one key position of the moving object according to the behavior; the at least one key position includes: at least one of a position where the velocity change amplitude is greater than a first amplitude threshold, a position where the acceleration change amplitude is greater than a second amplitude threshold, and a position where the direction angle change amplitude is greater than a third amplitude threshold; In the dynamic traffic data corresponding to the traffic trajectory, a data frame corresponding to the at least one key position is extracted to obtain a group of key frames corresponding to the traffic trajectory.

8. The method according to claim 6, characterized in that Extracting a data frame corresponding to the at least one key position from the dynamic traffic data corresponding to the traffic trajectory includes: For any key position of the at least one key position, obtaining, from the dynamic traffic data, a relative position relationship between the moving object and surrounding moving objects when the moving object is at the key position; According to the relative position relationship, determining whether the relative distance between the moving object and the surrounding moving objects is greater than a set distance threshold; If yes, then extract the data frame corresponding to the key position from the dynamic traffic data corresponding to the traffic trajectory.

9. The method according to any one of claims 1 to 5, characterized in that: Decision planning is performed according to the at least one set of key frames and preset constraints to obtain at least one driving trajectory, including: Under the constraint of the constraint condition, path planning and speed planning are performed according to the at least one group of key frames to obtain driving trajectories corresponding to the at least one group of key frames; The constraint conditions include at least one of the following: a deviation between a position of the driving trajectory at a key frame and a position perception result corresponding to the key frame is less than a set deviation threshold, a difference between a speed of the driving trajectory at the key frame and a speed perception result corresponding to the key frame is less than a set difference threshold, and the position of the driving trajectory is continuous, the speed of the driving trajectory is continuous, and the acceleration of the driving trajectory is continuous.

10. The method according to claim 9, characterized in that Also includes: adjusting the constraint conditions; Under the constraints of the adjusted constraints, path planning and speed planning are performed according to the at least one group of key frames, and the path planning results and / or speed planning results are disturbed by using preset perturbation parameters to obtain generalized driving trajectories corresponding to the at least one group of key frames.

11. An electronic device, characterized in that: include: Memory and processor; The memory is used to store one or more computer instructions; The processor is configured to execute the one or more computer instructions to: perform the steps in the method according to any one of claims 1-10.

12. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, it can implement the method for processing autonomous driving scene data as described in any one of claims 1 to 10.

Images