A method and apparatus for processing simulation test tasks.

By constructing simulated test scenarios for traffic intersections based on real road network data, and combining autonomous vehicle algorithm interfaces and scenario simulators, the problem of insufficient scenario coverage in existing autonomous driving algorithm testing is solved, achieving more comprehensive test results.

CN120686778BActive Publication Date: 2026-11-10BEIJING VEHICLE NETWORK TECH DEV CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202510773998.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-06-11
Publication Date
2026-11-10
Estimated Expiration
2045-06-11

AI Technical Summary

Technical Problem

In existing autonomous driving algorithm testing, the scenario coverage based on vehicle-to-road data collection is limited and the collection cost is high, while the realism of manually designed simulated data is not high, resulting in insufficient testing complexity and comprehensiveness, especially in the poor testing results at traffic intersections.

Method used

By creating simulated test scenarios for traffic intersections based on real road network data, and utilizing the characteristics of time continuity and traffic stress testing, continuous and stress scenarios are constructed. Simulation tests are then conducted by connecting to a scenario simulator through a vehicle algorithm interface. By combining random, long-range, and efficient vehicle selection rules, the comprehensiveness of the test is improved.

Benefits of technology

It enhances the realism, complexity, and diversity of test scenarios, expands the types and scope of test vehicles, and effectively improves the testing complexity and comprehensiveness of autonomous driving algorithms.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120686778B_ABST
    Figure CN120686778B_ABST
Patent Text Reader

Abstract

This invention relates to a method and apparatus for processing simulation test tasks. The method includes: receiving a test road / time period / mode; if it is a first mode, dividing the test time period into multiple consecutive intersection segment time periods and creating intersection test scenarios based on real road network data; if it is a second mode, creating intersection test scenarios based on road network data of the maximum pressure time period for each intersection within the test time period; and having a scenario simulator perform a simulation test once based on each intersection test scenario and output the corresponding test recording scenario; combining each test recording scenario with its corresponding vehicle object index and intersection test scenario to form a corresponding simulation test record, and combining all simulation test records to form a simulation test task report and saving it. This invention can effectively improve the testing complexity and comprehensiveness of autonomous driving algorithms.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and apparatus for processing simulation test tasks. Background Technology

[0002] The predictive-planning-decision-control algorithm (hereinafter referred to as the autonomous driving algorithm) of an autonomous driving system is used to predict the movement trajectories of other traffic participants around the vehicle based on perception information, plan the vehicle's driving trajectory based on the prediction results and high-precision maps, make decisions on the vehicle's traffic behavior type (such as going straight, changing lanes, overtaking, and stopping) based on the planned trajectory and traffic light status, and predict the vehicle's lateral / longitudinal driving control parameters based on the decision results and the vehicle's motion state, thereby outputting corresponding vehicle driving control commands (such as steering wheel angle, throttle opening degree, brake opening degree, gear position, etc.).

[0003] Simulation testing is a novel method for testing autonomous driving algorithms. It allows for scenario reproduction, improves testing efficiency, and reduces road testing costs. Currently, conventional simulation testing uses two types of raw data to construct scenarios: road-collected data from the vehicle's perspective and manually designed simulation data. Both types of data sources have shortcomings: 1) Vehicle-collected road-collected data has limited scenario coverage and high collection costs; 2) Manually designed data lacks realism and has weak scenario generalization. Testing autonomous driving algorithms solely based on these two types of data is insufficient in terms of complexity and comprehensiveness.

[0004] Traffic intersections are complex traffic scenarios, and the data on traffic networks is reliable and massive. Constructing simulation test scenarios based on real traffic network data at intersections would significantly improve the complexity and comprehensiveness of algorithm testing. Therefore, how to test autonomous driving algorithms using intersection simulation scenarios built from real road network data is precisely the technical problem this invention aims to solve. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by providing a method, apparatus, electronic device, and computer-readable storage medium for processing simulation testing tasks. This invention offers two methods for creating realistic intersection scenarios to enhance scenario complexity and diversity: continuous scenario creation for all intersections on any test road based on time continuity, and stress scenario creation for all intersections on any test road based on traffic stress testing characteristics. After scenario creation, simulation-testing is achieved by connecting the autonomous vehicle algorithm interface to the scenario simulator. Three types of autonomous vehicle selection rules (random, long-range, and efficient) are provided for each scenario test to improve the comprehensiveness of the test. The scenario simulator performs simulation testing of the autonomous driving algorithm based on the currently loaded intersection test scenario and through the connected autonomous vehicle algorithm interface, recording the real-time test scenario during the test. Based on this invention, the realism, complexity, and diversity of test scenarios can be enhanced, the types and range of test vehicles can be expanded, and the comprehensiveness of the test can be improved. This invention effectively improves the complexity and comprehensiveness of algorithm testing.

[0006] To achieve the above objectives, a first aspect of the present invention provides a method for processing simulation test tasks, the method comprising:

[0007] Receive test road R test Test period T test and test mode; and use the autonomous driving algorithm interface to be tested as the corresponding vehicle algorithm interface; the test road R test The test period T is a real traffic network road segment. test The timing period is 24 hours; the test mode includes a first mode and a second mode;

[0008] The test mode is identified; if the test mode is the first mode, then according to the test road R... test The real road network data will be used for the test period T test The road is divided into multiple consecutive intersection time segments, and test simulation scenarios are created for each intersection in its corresponding intersection time segment based on real road network data to obtain corresponding intersection test scenarios; if the test mode is the second mode, then according to the test road R test Real road network data for each intersection on the current road during the test period T test The corresponding intersection test scenario is obtained by creating a test simulation scenario during the period of maximum pressure within the time frame.

[0009] Each intersection test scenario is used as the corresponding current intersection test scenario; and a car object is selected from the current intersection test scenario as the corresponding autonomous vehicle object based on a preset autonomous vehicle selection rule; the current intersection test scenario is imported into a preset scene simulator for initialization; and the autonomous vehicle algorithm interface is connected to the scene simulator to take over the driving control of the autonomous vehicle object; and the scene simulator performs a simulation test according to the loaded current intersection test scenario and the connected autonomous vehicle algorithm interface and outputs the corresponding test recording scenario; the autonomous vehicle selection rule includes at least random, long-range, and high-efficiency;

[0010] The object index of the vehicle object in each of the test recording scenarios is used as the corresponding vehicle object index; and each of the test recording scenarios, its corresponding vehicle object index, and the intersection test scenario are combined to form a corresponding simulation test record; and all the obtained simulation test records are combined to form a corresponding simulation test task report and saved.

[0011] Preferably, the input data of the autonomous vehicle algorithm interface includes simulation timestamps, traffic participant datasets, traffic light datasets, autonomous vehicle status data, autonomous vehicle target position, and high-precision map data, and the output is vehicle driving control commands; the traffic participant dataset consists of one or more traffic participant data, which includes participant type, participant coordinates, participant speed, participant orientation angle, and participant acceleration; the traffic light dataset consists of one or more traffic light data, which includes traffic light coordinates and lighting status, including red light, green light, yellow light, and flashing yellow light; the autonomous vehicle status data includes autonomous vehicle coordinates, autonomous vehicle orientation angle, autonomous vehicle speed, and autonomous vehicle acceleration; the autonomous vehicle target position is a target position coordinate; the command parameters of the vehicle driving control commands include at least steering wheel angle, accelerator pedal opening degree, brake pedal opening degree, and gear position;

[0012] The test road R test Including multiple intersections C i and multiple road sections j 1 ≤ intersection index i ≤ N, 1 ≤ road segment index j ≤ N-1, where N is the total number of intersections; the test road R test The starting and ending positions are at intersection C. i=1 C i=N The road segment r j Intersection C i=j C i=j+1 Connecting road sections between them;

[0013] The test period T test The start and end times are denoted as t. start t end ;

[0014] The test period T test Test duration L test =t end -t start ;

[0015] Each of the aforementioned intersections C i The corresponding intersection time segment is denoted as TA. i TA i ∈T test The two adjacent intersection time slots TA before and after. i TA i+1 In this process, the end time of the preceding time period is aligned with the start time of the following time period; the intersection segmentation time TA i The start and end times are denoted as ta s,i ,ta e,i ,ta s,i=1 =t start ,ta s,i+1 =ta e,i ,ta e,i=N =t end ;

[0016] Each of the aforementioned intersections C i The corresponding period of maximum pressure is denoted as TB. i TB i ∈T test The maximum pressure period TB i The start and end times are denoted as tb. s,i tb e,i ;

[0017] Each of the aforementioned intersections C i The corresponding intersection test scenario and the corresponding test recording scenario are denoted as the corresponding The test recording scenario With the intersection test scenario One-to-one correspondence; the intersection test scenario The test recording scenario The data encapsulation format all conforms to the OpenSCENARIO standard scene file data encapsulation format;

[0018] The intersection test scenario The data content includes at least a high-precision map of the intersection. and scene object set The high-precision map of the intersection The map elements include at least the traffic signs / markers / markings elements, road edge line elements, danger zone elements, obstacle elements, and traffic light elements for all roads at the current intersection; the scene object set Includes multiple scene objects 1 ≤ object index o ≤ N D,i N D,i The total number of objects in the test scenario at the i-th intersection; the scenario objects Including object types Object state trajectory t is a timestamp; the object type At least including cars, motorcycles, bicycles, pedestrians, and traffic lights; the state trajectories of all described objects. The trajectory sampling frequency 1 / Δt remains consistent; the object type When representing cars, motorcycles, bicycles, and pedestrians, the object state is... It includes at least coordinates, orientation angle, velocity, and acceleration; the object type When it is a traffic light, the state of the object This includes the coordinates of the traffic lights and the status of the traffic lights, wherein the status of the traffic lights includes at least red light, green light, yellow light, and flashing yellow light;

[0019] The test recording scenario The data content includes at least a high-precision map of the intersection. and recording object set The high-precision map of the intersection The corresponding high-precision map of the intersection Maintain consistency; the set of recorded objects Including N D,i Recording objects The recording object With the scene object One-to-one correspondence; the recorded object Including object types and object state trajectory The object type With the corresponding object type Maintain consistency; the object's state trajectory The corresponding object state trajectory Timestamp alignment; the object type When the object is a car, motorcycle, bicycle, or pedestrian, the object state is as follows: It includes at least coordinates, orientation angle, velocity, and acceleration; the object type When it is a traffic light, the state of the object This includes the traffic light coordinates and the traffic light status, wherein the traffic light status includes at least red light, green light, yellow light, and flashing yellow light;

[0020] The scene simulator is a type of scene simulation tool that conforms to the OpenSCENARIO standard.

[0021] Preferably, the step based on the test road R test The real road network data will be used for the test period T test The time segment is divided into multiple consecutive intersection segments, specifically including:

[0022] Based on the test road R test The real road network data is used to analyze each of the aforementioned road segments r within a preset specified time period. j During the test period T test The average vehicle speed within the area is statistically analyzed to obtain the corresponding average vehicle speed. and each of the aforementioned road segments r j The length of the straight road is denoted as the corresponding length d. j ; and based on each of the aforementioned lengths d j and the corresponding average vehicle speed Calculate the corresponding average travel time And for N-1 of the average travel times Δt j The total duration is calculated by summing the total duration. sum The specified time period includes at least a specified single day or multiple consecutive days, a specified single week or multiple consecutive weeks, a specified single month or multiple consecutive months, a specified single quarter or multiple consecutive quarters, or a specified single year or multiple consecutive years.

[0023] And the total duration l sum With the test duration L test Perform identification; if the total duration is l sum Less than the test duration L test Then set the corresponding end intersection segment duration △l end =L test -l sum If the total duration is l sum Greater than or equal to the test duration L test Then, any one of the average travel times Δl can be selected. j As the corresponding incremental duration △l add and the test period T test The end time t end The test duration L was postponed to make the latest test duration L test Satisfy L test =l sum +△l add And set the corresponding segmentation time Δl at the tail intersection. end =△l add ;

[0024] And based on the obtained N-1 average travel times Δl jThe segmentation time Δl at the tail intersection end and the test period T test The start time t start and the end time t end For the N intersection time segments TA i The start time ta s,i and the end time ta e,i Configure settings:

[0025] ta s,i=1 =t start ,ta e,i=1 =ta s,i=1 +Δl j=i ,

[0026] ta s,1<i≤N-1 =ta e,i-1 ,ta e,1<i≤N-1 =ta s,1<i≤N-1 +Δl j=i ,

[0027] ta s,i=N =ta e,i=N-1 ,ta s,i=N =ta s,i=N +Δl end =t end .

[0028] Preferably, the step of creating corresponding intersection test scenarios based on real road network data for each intersection in its corresponding intersection time segment to obtain the corresponding intersection test scenario specifically includes:

[0029] Based on the test road R test The real road network data is used to measure the current road during the pre-set specified time period T each day. test The total number of traffic flows is counted to obtain the corresponding first total, and the single-day road network data corresponding to the largest first total is used as the road network data for that day.

[0030] And the road network data of the day and each of the intersections C i The corresponding regional road network data is used as the corresponding intersection area data; and the intersection area data is divided into the corresponding intersection time slots (TA). i The time-period road network data is used as the corresponding intersection time-period data;

[0031] and each of the aforementioned intersections C i The corresponding high-precision map serves as the corresponding high-precision map of the intersection. and each of the aforementioned intersections C iThe corresponding intersection time period data is used as the corresponding current intersection time period data; and the motion trajectory of each car, motorcycle, bicycle or pedestrian in the current intersection time period data is sampled according to a preset trajectory sampling frequency, and the corresponding scene object is constructed based on the sampling results. The coordinates and changing trajectories of each traffic light in the current intersection time period data are sampled according to the trajectory sampling frequency, and the corresponding scene object is constructed based on the sampling results. And from all the scene objects obtained Form the corresponding scene object set And from the obtained high-precision map of the intersection and the set of scene objects The corresponding intersection test scenario is formed.

[0032] Preferably, the step based on the test road R test Real road network data for each intersection on the current road during the test period T test The test simulation scenario during the period of maximum pressure within the time frame is used to create the corresponding intersection test scenario, specifically including:

[0033] Each of the aforementioned intersections C i As the corresponding current intersection; and based on the test road R test The real road network data is used to test the current intersection during the specified test period T each day within a preset time period. test The maximum number of traffic participants within the area is used to calculate the corresponding second total; and the largest second total is denoted as the total y. max,i ; and the total number y max,i The corresponding daily road network data is related to the current intersection and the test period T. test The corresponding regional time period road network data is used as the corresponding current intersection data; and the total number y is used as the data. max,i During the test period T test The corresponding time in the table is taken as the corresponding maximum pressure period TB. i The start time tb s,i and the test period T test The end time t end The current maximum pressure period TB i The end time tb e,i ; and the current intersection data that meets the current maximum pressure period TB i The road network data is used as the corresponding intersection pressure period data;

[0034] and each of the aforementioned intersections C iThe corresponding high-precision map serves as the corresponding high-precision map of the intersection. and each of the aforementioned intersections C i The corresponding intersection pressure time period data is used as the corresponding current pressure time period data; and the motion trajectory of each car, motorcycle, bicycle or pedestrian in the current pressure time period data is sampled according to a preset trajectory sampling frequency, and the corresponding scene object is constructed based on the sampling results. The coordinates and changing trajectories of each traffic light in the current pressure period data are sampled according to the trajectory sampling frequency, and the corresponding scene object is constructed based on the sampling results. And from all the scene objects obtained Form the corresponding scene object set And from the obtained high-precision map of the intersection and the set of scene objects The corresponding intersection test scenario is formed.

[0035] Preferably, the step of selecting a car object from the current intersection test scenario based on a preset vehicle selection rule as the corresponding vehicle object specifically includes:

[0036] The scene object set of the current intersection test scenario. The object type described in The scene object for the car The objects are clustered into a corresponding set of car objects; the vehicle selection rule is identified; if the vehicle selection rule is random, a scene object is randomly selected from the set of car objects. As the corresponding vehicle object; if the vehicle selection rule is long-range, then the object state trajectory in the vehicle object set. The longest of the scene objects As the corresponding vehicle object; if the vehicle selection rule is efficient, then the object state trajectory in the vehicle object set. The scene object with the fastest average speed As the corresponding vehicle object.

[0037] Preferably, the step of importing the current intersection test scenario into a preset scenario simulator for initialization specifically includes:

[0038] The scene simulator uses the high-precision map of the intersection in the currently imported test scene. Construct a two-dimensional or three-dimensional simulation world; and based on the scene object set of the current intersection test scenario. Each of the aforementioned scene objects In the current simulation world, a one-to-one simulation object is constructed; the simulation time period of the current simulation world is configured based on the time span of the current intersection test scenario; the recording sampling frequency and single-step simulation frequency of the current simulation world are configured based on the trajectory sampling frequency of the current intersection test scenario, wherein the recording sampling frequency is consistent with the current trajectory sampling frequency, and the single-step simulation frequency is an integer multiple of the current trajectory sampling frequency; simulation objects of type car, motorcycle, bicycle, or pedestrian in the current simulation world are regarded as participant objects, and corresponding traffic participant behavior models, motion control models, and dynamic models are assigned to each participant object; simulation objects of type traffic light are regarded as traffic light objects, and corresponding traffic light control models are assigned to each traffic light object; and the simulation is based on the object state trajectory corresponding to each simulation object. The first object state in The initial simulation state of the current simulation object is set, and the second to last object states of each simulation object are set. The multi-stage simulation objective serves as the current simulation object;

[0039] The traffic participant behavior model is used to determine the traffic behavior type of the current participant at a future simulation time based on the current simulation goal and current historical trajectory of each participant at the current simulation time, the historical trajectories of other participants at the current simulation time, and the historical trajectories of surrounding traffic light objects at the current simulation time. The traffic behavior types include at least decelerating straight ahead, accelerating straight ahead, driving straight ahead at a constant speed, changing lanes to the left, changing lanes to the right, overtaking, making a U-turn, yielding to the left, yielding to the right, sudden braking, and stopping.

[0040] The motion control model is used to predict the motion control parameters of each participant at a future simulation time based on the current simulation goal, current traffic behavior type, and current historical trajectory at the current simulation moment; the motion control parameters include at least steering force parameters, longitudinal acceleration force parameters, and longitudinal deceleration force parameters;

[0041] The dynamic model is used to predict the simulation state of the current participant at future simulation moments based on the current simulation target, current simulation state, and current motion control parameters of each participant at the current simulation moment; the simulation state includes at least coordinates, orientation angle, velocity, and acceleration;

[0042] The traffic light control model is used to predict the traffic light state of the current traffic light object at future simulation times based on the current simulation target and current historical trajectory of each traffic light object at the current simulation time.

[0043] Preferably, the step of connecting the autonomous vehicle algorithm interface to the scene simulator to take over the driving control of the autonomous vehicle specifically includes:

[0044] The simulation object corresponding to the autonomous vehicle in the current simulation world of the scene simulator is taken as the corresponding current takeover object; the traffic participant behavior model and motion control model assigned to the current takeover object by the scene simulator are canceled; the input end of the dynamics model assigned to the current takeover object by the scene simulator is connected to the output end of the autonomous vehicle algorithm interface; and a corresponding simulation data output interface is set in the scene simulator and connected to the input end of the autonomous vehicle algorithm interface.

[0045] The simulation data output interface is used to, at each simulation moment in the simulation process, take the real-time simulation state of the currently taking over object in the current simulation world as the corresponding vehicle state data; take the real-time simulation state of each of the other participating objects around the currently taking over object as the corresponding traffic participant data, and form a corresponding traffic participant dataset from all the obtained traffic participant data; take the coordinates and real-time traffic light state of each traffic light object in the current simulation world as the corresponding traffic light coordinates and lighting state to form a corresponding traffic light data, and form a corresponding traffic light dataset from all the obtained traffic light data; take the coordinate position of the most recent next-stage simulation target in the multi-stage simulation target corresponding to the currently taking over object as the corresponding vehicle target position; and take the high-precision map of the intersection corresponding to the current simulation world. The high-precision map data is used as the corresponding data; the current simulation time is used as the corresponding simulation timestamp; and the obtained simulation timestamp, traffic participant dataset, traffic light dataset, vehicle status data, vehicle target location, and high-precision map data are used as the output data for this simulation and output to the vehicle algorithm interface.

[0046] Preferably, the step of the scene simulator performing a simulation test based on the loaded current intersection test scenario and the accessed autonomous vehicle algorithm interface, and outputting the corresponding test recording scenario, specifically includes:

[0047] During the simulation process, the scene simulator simulates the motion trajectory of the current participant object using the traffic participant behavior model, motion control model, and dynamics model of each participant object; it also simulates the changing trajectory of the current traffic light object using the traffic light control model of each traffic light object; and it simulates the motion trajectory of the vehicle object using the simulation data output interface, vehicle algorithm interface, and dynamics model corresponding to the vehicle object; and based on the recording sampling frequency, it collects data on the real-time simulation state of each simulation object at each recording sampling point to generate the corresponding object state. At the end of the simulation, the high-precision map of the intersection in the current intersection test scenario will be generated. As the corresponding high-precision map of the intersection and the object type corresponding to each of the simulation objects As a corresponding object type and the states of all objects corresponding to each of the simulation objects. Form a corresponding object state trajectory and the object type corresponding to each of the simulation objects. and the object state trajectory Form a corresponding recording object And by all the recorded objects The corresponding set of recording objects is formed. And from the high-precision map of the intersection and the set of recorded objects The corresponding test recording scenario is formed. And output it.

[0048] A second aspect of the present invention provides an apparatus for implementing the simulation test task processing method described in the first aspect above, the apparatus comprising: a test task receiving module, an intersection scene preparation module, a simulation test module, and a task report saving module;

[0049] The test task receiving module is used to receive test road R test Test period T test and test mode; and use the autonomous driving algorithm interface to be tested as the corresponding vehicle algorithm interface; the test road R test The test period T is a real traffic network road segment. test The timing period is 24 hours; the test mode includes a first mode and a second mode;

[0050] The intersection scene preparation module is used to identify the test mode; if the test mode is the first mode, then according to the test road R test The real road network data will be used for the test period T test The road is divided into multiple consecutive intersection time segments, and test simulation scenarios are created for each intersection in its corresponding intersection time segment based on real road network data to obtain corresponding intersection test scenarios; if the test mode is the second mode, then according to the test road R test Real road network data for each intersection on the current road during the test period T test The corresponding intersection test scenario is obtained by creating a test simulation scenario during the period of maximum pressure within the time frame.

[0051] The simulation testing module is used to take each intersection test scenario as the corresponding current intersection test scenario; and select a car object from the current intersection test scenario as the corresponding autonomous vehicle object based on a preset autonomous vehicle selection rule; import the current intersection test scenario into a preset scene simulator for initialization; and connect the autonomous vehicle algorithm interface to the scene simulator to take over the driving control of the autonomous vehicle object; and the scene simulator performs a simulation test according to the loaded current intersection test scenario and the connected autonomous vehicle algorithm interface and outputs the corresponding test recording scenario; the autonomous vehicle selection rule includes at least random, long-range, and high-efficiency;

[0052] The task report saving module is used to use the object index of the vehicle object in each of the test recording scenarios as the corresponding vehicle object index; and to form a corresponding simulation test record by combining each of the test recording scenarios, its corresponding vehicle object index, and the intersection test scenario; and to form a corresponding simulation test task report by combining all the obtained simulation test records and saving it.

[0053] A third aspect of the present invention provides an electronic device, including: a memory, a processor, and a transceiver;

[0054] The processor is used to couple with the memory, read and execute instructions in the memory to implement the steps of the method described in the first aspect above;

[0055] The transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.

[0056] A fourth aspect of the present invention provides a computer-readable storage medium storing computer instructions that, when executed by a computer, cause the computer to perform the instructions described in the first aspect.

[0057] This invention provides a method, apparatus, electronic device, and computer-readable storage medium for processing simulation test tasks. As described above, this invention provides two methods for creating realistic intersection scenarios to enhance scenario complexity and diversity: continuous scenario creation for all intersections on any test road based on time continuity, and stress scenario creation for all intersections on any test road based on traffic stress test characteristics. After scenario creation, simulation-testing is achieved by connecting the autonomous vehicle algorithm interface to the scenario simulator. Three types of autonomous vehicle selection rules (random, long-range, and efficient) are provided for each scenario test to improve the comprehensiveness of the test. The scenario simulator performs simulation testing of the autonomous driving algorithm based on the currently loaded intersection test scenario and through the connected autonomous vehicle algorithm interface, and records the real-time test scenario during the test. This invention enhances the realism, complexity, and diversity of test scenarios, expands the types and range of test vehicles, and effectively improves the complexity and comprehensiveness of algorithm testing. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of a simulation test task processing method provided in Embodiment 1 of the present invention;

[0059] Figure 2 This is a schematic diagram of the test road, intersection, and road section provided in Embodiment 1 of the present invention;

[0060] Figure 3 This is a schematic diagram of intersection time segmentation provided in Embodiment 1 of the present invention;

[0061] Figure 4 This is a schematic diagram of the maximum pressure period provided in Embodiment 1 of the present invention;

[0062] Figure 5 This is a module structure diagram of a simulation test task processing device provided in Embodiment 2 of the present invention;

[0063] Figure 6 This is a schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present invention. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0065] Embodiment 1 of the present invention provides a method for processing simulation test tasks, such as... Figure 1The schematic diagram of a simulation test task processing method provided in Embodiment 1 of the present invention is shown, which includes the following main steps:

[0066] Step 1, Receive test road R test Test period T test And test modes; and use the autonomous driving algorithm interface to be tested as the corresponding autonomous vehicle algorithm interface.

[0067] Here, the test road R in this embodiment of the invention test This is a real traffic network road; the test road R test Including multiple intersections C i and multiple road sections j 1 ≤ intersection index i ≤ N, 1 ≤ road segment index j ≤ N-1, N is the total number of intersections; test road R test The starting and ending positions are at intersection C. i=1 C i=N Road segment r j Intersection C i=j C i=j+1 The connecting road sections between them. Test road R test Intersection C i and road section r j Subordination and connection relationships, such as Figure 2 The diagram shows the test road, intersection, and road segment provided in Embodiment 1 of the present invention. The test period T of this embodiment of the present invention... test The timing period is 24 hours; the test period T test The start and end times are denoted as t. start t end Test period T test Test duration L test =t end -t start .

[0068] The testing modes of this invention include a first mode and a second mode. In the first mode, continuous scenarios are created for all intersections of any test road based on real road network data and according to time continuity. In the second mode, stress scenarios are created for all intersections of any test road based on real road network data and according to the characteristics of traffic stress testing. Real road network data refers to all-weather, full-segment / intersection traffic data collected by roadside edge devices of a real traffic network, including at least the latest high-precision map of the current road segment / intersection, the movement trajectory information of all traffic participants in the current road segment / intersection, the changing trajectory information of all traffic lights in the current road segment / intersection, and information on all construction / accident / dangerous areas in the current road segment / intersection, etc.

[0069] The autonomous vehicle algorithm interface of this invention is the interface for any automaker's autonomous driving algorithm / model / device / system. The input data of this autonomous vehicle algorithm interface includes simulation timestamps, traffic participant datasets, traffic light datasets, autonomous vehicle status data, autonomous vehicle target location, and high-precision map data. The output is vehicle driving control commands. Specifically: 1) The traffic participant dataset consists of one or more traffic participant data sets, including participant type, participant coordinates, participant speed, participant orientation angle, and participant acceleration; 2) The traffic light dataset consists of one or more traffic light data sets, including traffic light coordinates and illumination status, including red, green, yellow, and flashing yellow; 3) The autonomous vehicle status data includes autonomous vehicle coordinates, autonomous vehicle orientation angle, autonomous vehicle speed, and autonomous vehicle acceleration; 4) The autonomous vehicle target location is a target location coordinate; 5) The command parameters of the vehicle driving control commands include at least steering wheel angle, throttle opening, brake opening, and gear position.

[0070] Step 2: Identify the test mode; if the test mode is the first mode, then based on the test road R... test The real road network data will be used for the test period T test The road is divided into multiple consecutive intersection time segments, and test simulation scenarios are created for each intersection in its corresponding intersection time segment based on real road network data to obtain the corresponding intersection test scenarios; if the test mode is the second mode, then according to the test road R test Real road network data for each intersection on the current road during the test period T test The corresponding intersection test scenario is obtained by creating a test simulation scenario during the period of maximum pressure within the time frame.

[0071] Specifically, this includes: Step 21, identifying the test pattern;

[0072] Step 22, if the test mode is the first mode, then according to the test road R test The real road network data will be used for the test period T test The road is divided into multiple consecutive intersection time segments, and test simulation scenarios are created for each intersection in its corresponding intersection time segment based on real road network data to obtain the corresponding intersection test scenarios.

[0073] Here, in the embodiments of the present invention, each intersection C i The corresponding intersection time segment is denoted as TA. i TA i ∈T test ; Time slots for the two adjacent intersections before and after i TA i+1 In the middle, the end time of the previous period is aligned with the start time of the subsequent period; intersection segmented time TA iThe start and end times are denoted as ta s,i ,ta e,i ,ta s,i=1 =t start ,ta s,i+1 =ta e,i ,ta e,i=N =t end ;like Figure 3 The diagram shows the intersection segmentation time period provided in Embodiment 1 of the present invention;

[0074] Each intersection C in the embodiments of the present invention i The corresponding intersection test scenario is denoted as Intersection test scenario The data encapsulation format conforms to the OpenSCENARIO standard scene file data encapsulation format; intersection test scenario The data content includes at least a high-precision map of the intersection. and scene object set High-precision map of intersection The map elements should include at least the traffic signs / markers / markings elements, road edge line elements, danger zone elements, obstacle elements, and traffic light elements for all roads at the current intersection; scene object set. Includes multiple scene objects 1 ≤ object index o ≤ N D,i N D,i The total number of objects in the test scenario at the i-th intersection; scenario objects Including object types Object state trajectory t is a timestamp; object type Includes at least cars, motorcycles, bicycles, pedestrians, and traffic lights; and the state trajectories of all objects. The trajectory sampling frequency 1 / Δt remains consistent; object type When representing cars, motorcycles, bicycles, and pedestrians, the object state is... At least include coordinates, orientation angle, velocity, and acceleration; object type When it is a traffic light, the object state This includes traffic light coordinates and traffic light status, with traffic light status including at least red, green, yellow, and flashing yellow.

[0075] The current step 22 specifically includes:

[0076] Step 221, based on test road R test The real road network data will be used for the test period T test Divide the time into multiple consecutive intersection segments;

[0077] Specifically, this includes: Step 2211, based on test road R test The real road network data is used to analyze the road segments r within a preset specified time period. j During test period T test The average vehicle speed within the area is statistically analyzed to obtain the corresponding average vehicle speed. and each road section r j The length of the straight road is denoted as the corresponding length d. j ; and based on each length d j and their corresponding average speed Calculate the corresponding average travel time And for N-1 average travel times Δt j The total duration is calculated by summing the total duration. sum ;

[0078] Here, the preset specified time period mentioned in the embodiments of the present invention includes at least a specified single day or multiple consecutive days, a specified single week or multiple consecutive weeks, a specified single month or multiple consecutive months, a specified single quarter or multiple consecutive quarters, and a specified single year or multiple consecutive years.

[0079] Step 2212, and calculate the total duration l sum With test duration L test Perform identification; if the total duration is l sum Less than test duration L test Then set the corresponding end intersection segment duration △l end =L test -l sum If the total duration is l sum Greater than or equal to test duration L test Then, choose any average travel time Δl j As the corresponding incremental duration △l add and the test period T test End time t end The test duration L was postponed. test Satisfy L test =l sum +△l add And set the corresponding end intersection segment duration △l end =△l add ;

[0080] Step 2213, and based on the obtained N-1 average travel times Δl j End intersection segmentation time △l end and the test period T test start time t start and end time t end For N intersections, segment the time period TAi The start time ta s,i and end time ta e,i Configure settings:

[0081] ta s,i=1 =t start ,ta e,i=1 =ta s,i=1 +Δl j=i ,

[0082] ta s,1<i≤N-1 =ta e,i-1 ,ta e,1<i≤N-1 =ta s,1<i≤N-1 +Δl j=i ,

[0083] ta s,i=N =ta e,i=N-1 ,ta s,i=N =ta s,i=N +Δl end =t end ;

[0084] Step 222: Based on real road network data, create test simulation scenarios for each intersection of the current road in its corresponding intersection segment time period to obtain the corresponding intersection test scenarios;

[0085] Specifically, this includes: Step 2221, based on test road R test The real road network data is used to test the current road during the daily test period T within a preset specified time period. test The total number of vehicles on the road is counted to obtain the corresponding first total, and the single-day road network data corresponding to the largest first total is used as the road network data for that day.

[0086] Step 2222, and then compare the road network data of the day with the data from each intersection C. i The corresponding regional road network data is used as the corresponding intersection area data; and the intersection area data is divided into corresponding intersection time slots (TA). i The time-period road network data is used as the corresponding intersection time-period data;

[0087] Step 2223, and then at each intersection C i The corresponding high-precision map serves as the corresponding high-precision map for the intersection. and each intersection C i The corresponding intersection time period data is used as the corresponding current intersection time period data; and the motion trajectory of each car, motorcycle, bicycle or pedestrian in the current intersection time period data is sampled according to the preset trajectory sampling frequency, and the corresponding scene object is constructed based on the sampling results. The system samples the coordinates and changing trajectories of each traffic light in the current intersection time period data according to the trajectory sampling frequency, and constructs corresponding scene objects based on the sampling results. And from all the scene objects obtained Form the corresponding scene object set And the high-precision map of the intersection obtained and scene object set Form corresponding intersection test scenarios

[0088] Here, the preset trajectory sampling frequency is the object state trajectory. The trajectory sampling frequency, which can be preset or dynamically configured based on application requirements;

[0089] Step 23, if the test mode is the second mode, then according to the test road R test Real road network data for each intersection on the current road during the test period T test The corresponding intersection test scenario is obtained by creating a test simulation scenario during the period of maximum pressure within the time frame.

[0090] Among them, at each intersection C i The corresponding period of maximum pressure is denoted as TB. i TB i ∈T test ; Maximum pressure period TB i The start and end times are denoted as tb. s,i tb e,i ;

[0091] Specifically, this includes: Step 231, which involves each intersection C... i As the corresponding current intersection; and based on the test road R test Based on real road network data, the test period T is used for the current intersection during a pre-set specified time period each day. test The maximum number of traffic participants within the area is used to calculate the corresponding second total; and the largest second total is denoted as the total y. max,i ; and the total number y max,i The corresponding daily road network data is related to the current intersection and the test period T. test The corresponding regional time period road network data is used as the corresponding current intersection data; and the total number y is... max,i During test period T test The corresponding time in the table is taken as the corresponding period of maximum pressure (TB). i start time tb s,i and the test period T test End time t end TB is currently experiencing its highest pressure period. i End time tb e,i; and will include the current intersection data that meets the current maximum pressure period TB i The road network data is used as the corresponding intersection pressure period data;

[0092] Here, the period of maximum pressure is TB i With the total number y max,i and test period T test Relationship such as Figure 4 The diagram shows the maximum pressure period provided in Embodiment 1 of the present invention;

[0093] Step 232, and set each intersection C i The corresponding high-precision map serves as the corresponding high-precision map for the intersection. and each intersection C i The corresponding intersection traffic pressure time period data is used as the corresponding current traffic pressure time period data; and the motion trajectory of each car, motorcycle, bicycle or pedestrian in the current traffic pressure time period data is sampled according to the preset trajectory sampling frequency, and the corresponding scene object is constructed based on the sampling results. The system samples the coordinates and changing trajectories of each traffic light in the current pressure period data based on the trajectory sampling frequency, and constructs corresponding scene objects based on the sampling results. And from all the scene objects obtained Form the corresponding scene object set And the high-precision map of the intersection obtained and scene object set Form corresponding intersection test scenarios

[0094] Step 3: Take each intersection test scenario as the corresponding current intersection test scenario; select a car object from the current intersection test scenario as the corresponding autonomous vehicle object based on the preset autonomous vehicle selection rules; import the current intersection test scenario into the preset scene simulator for initialization; connect the autonomous vehicle algorithm interface to the scene simulator to take over the driving control of the autonomous vehicle object; and have the scene simulator perform a simulation test according to the loaded current intersection test scenario and the connected autonomous vehicle algorithm interface and output the corresponding test recording scenario.

[0095] Specifically, this includes: Step 31, taking each intersection test scenario as the corresponding current intersection test scenario;

[0096] Step 32, and select a car object from the current intersection test scenario as the corresponding autonomous vehicle object based on the preset autonomous vehicle selection rules; wherein, the autonomous vehicle selection rules include at least random, long-range and efficient.

[0097] Specifically, this includes: setting the scene object set for the current intersection test scenario. object type Scene objects for cars The objects are clustered into a corresponding set of car objects; the vehicle selection rule is identified; if the vehicle selection rule is random, a scene object is randomly selected from the set of car objects. As the corresponding vehicle object; if the vehicle selection rule is long-range, then the state trajectory of the vehicle object set will be... Longest scene object As the corresponding vehicle object; if the vehicle selection rule is efficient, then the vehicle object set will be used to determine the object state trajectory. The fastest average speed scene object As the corresponding vehicle object;

[0098] Step 33, and import the current intersection test scenario into the preset scenario simulator for initialization;

[0099] Specifically, this includes: the scene simulator using a high-precision map of the current intersection test scene imported from the current network. Construct a 2D or 3D simulation world; and based on the scene object set of the current intersection test scenario. Various scene objects In the current simulation world, a one-to-one simulation object is constructed; the simulation time period of the current simulation world is configured based on the time span of the current intersection test scenario; the recording sampling frequency and single-step simulation frequency of the current simulation world are configured based on the trajectory sampling frequency of the current intersection test scenario, with the recording sampling frequency consistent with the current trajectory sampling frequency, and the single-step simulation frequency being an integer multiple of the current trajectory sampling frequency; simulation objects of type car, motorcycle, bicycle, or pedestrian in the current simulation world are regarded as participant objects, and corresponding traffic participant behavior models, motion control models, and dynamic models are assigned to each participant object; simulation objects of type traffic light are regarded as traffic light objects, and corresponding traffic light control models are assigned to each traffic light object; and the simulation is based on the object state trajectory of each simulation object. The first object state in Set the initial simulation state of the current simulation object, and set the second to last object states of each simulation object. The multi-stage simulation objective serves as the current simulation object;

[0100] Here, the scene simulator in this embodiment of the invention is a scene simulation tool that meets the OpenSCENARIO standard;

[0101] In this embodiment of the invention, the time span of the current intersection test scenario is the intersection segment time period (TA) corresponding to the current intersection test scenario. i or the period of maximum pressure TB i;

[0102] The traffic participant behavior model in this embodiment of the invention is used to make decisions on the traffic behavior type of the current participant at a future simulation time based on the current simulation goal and current historical trajectory of each participant at the current simulation time, the historical trajectories of other participants at the current simulation time, and the historical trajectories of traffic lights at the current simulation time. The traffic behavior types include at least decelerating straight ahead, accelerating straight ahead, driving straight ahead at a constant speed, changing lanes to the left, changing lanes to the right, overtaking, making a U-turn, yielding to the left, yielding to the right, emergency braking, and stopping.

[0103] The motion control model in this embodiment of the invention is used to predict the motion control parameters of each participant at a future simulation time based on the current simulation goal, current traffic behavior type, and current historical trajectory at the current simulation moment; the motion control parameters include at least steering force parameters, longitudinal acceleration force parameters, and longitudinal deceleration force parameters;

[0104] The dynamic model in this embodiment of the invention is used to predict the simulation state of the current participant at a future simulation time based on the current simulation target, current simulation state, and current motion control parameters of each participant at the current simulation moment; the simulation state includes at least coordinates, orientation angle, velocity, and acceleration;

[0105] The traffic light control model of this invention is used to predict the traffic light state of the current traffic light object at future simulation times based on the current simulation target and current historical trajectory of each traffic light object at the current simulation time.

[0106] Step 34, and connect the autonomous vehicle algorithm interface to the scene simulator to take over the driving control of the autonomous vehicle object;

[0107] Specifically, this includes: taking the simulation object corresponding to the vehicle object in the current simulation world of the scene simulator as the corresponding current takeover object; canceling the traffic participant behavior model and motion control model assigned to the current takeover object by the scene simulator; connecting the input end of the dynamics model assigned to the current takeover object by the scene simulator to the output end of the vehicle algorithm interface; and setting up a corresponding simulation data output interface in the scene simulator to connect to the input end of the vehicle algorithm interface.

[0108] Here, the simulation data output interface of this embodiment of the invention is used to take the real-time simulation state of the currently taking over object in the current simulation world as the corresponding vehicle state data at each simulation moment in the simulation process; and to take the real-time simulation state of each other participant object around the currently taking over object as the corresponding traffic participant data, and to form a corresponding traffic participant dataset from all the obtained traffic participant data; and to take the coordinates and real-time traffic light state of each traffic light object in the current simulation world as the corresponding traffic light coordinates and lighting state to form a corresponding traffic light data, and to form a corresponding traffic light dataset from all the obtained traffic light data; and to take the coordinate position of the most recent next-stage simulation target in the multi-stage simulation target corresponding to the currently taking over object as the corresponding vehicle target position; and to take the high-precision map of the intersection corresponding to the current simulation world. The high-precision map data is used as the corresponding simulation time; the current simulation time is used as the corresponding simulation timestamp; and the obtained simulation timestamp, traffic participant dataset, traffic light dataset, vehicle status data, vehicle target location and high-precision map data are used as the output data for this time and output to the vehicle algorithm interface.

[0109] Step 35, and the scene simulator performs a simulation test based on the loaded current intersection test scene and the accessed autonomous vehicle algorithm interface, and outputs the corresponding test recording scene;

[0110] Here, in the embodiments of the present invention, each intersection C i The corresponding test recording scenario is denoted as Test recording scenario Intersection test scenario One-to-one correspondence; test recording scenario The data encapsulation format conforms to the OpenSCENARIO standard scene file data encapsulation format; test recording scene The data content includes at least a high-precision map of the intersection. and recording object set High-precision map of intersection High-precision map of the corresponding intersection Maintain consistency; record object set Including N D,i Recording objects Recording object With scene objects One-to-one correspondence; recording object Including object types and object state trajectory object type With the corresponding object type Maintain consistency; object state trajectory Corresponding object state trajectory Timestamp alignment; object type When representing cars, motorcycles, bicycles, and pedestrians, the object state is... At least include coordinates, orientation angle, velocity, and acceleration; object type When it is a traffic light, the object state This includes traffic light coordinates and traffic light status, with traffic light status including at least red, green, yellow, and flashing yellow.

[0111] The current step 35 specifically includes: during the simulation process, the scene simulator simulates the motion trajectory of the current participant object using the traffic participant behavior model, motion control model, and dynamics model of each participant object; simulates the changing trajectory of the current traffic light object using the traffic light control model of each traffic light object; simulates the motion trajectory of the vehicle object using the simulation data output interface, vehicle algorithm interface, and dynamics model corresponding to the vehicle object; and, based on the recording sampling frequency, collects data on the real-time simulation state of each simulation object at each recording sampling point to generate the corresponding object state. At the end of the simulation, a high-precision map of the intersection in the current test scenario will be generated. As the corresponding high-precision map of the intersection and the object type corresponding to each simulation object As a corresponding object type And the state of all objects corresponding to each simulation object. Form a corresponding object state trajectory And by the object type corresponding to each simulation object. and object state trajectory Form a corresponding recording object And by all recorded objects Form the corresponding recording object set And from the high-precision map of the intersection and recording object set Form corresponding test recording scenarios And output it.

[0112] Step 4: Use the object index of the vehicle object in each test recording scenario as the corresponding vehicle object index; and combine each test recording scenario with its corresponding vehicle object index and the intersection test scenario to form a corresponding simulation test record; and combine all the obtained simulation test records to form a corresponding simulation test task report and save it.

[0113] It should be noted that the method in this embodiment of the invention further includes: after obtaining the simulation test task report, comprehensively evaluating the safety, efficiency, comfort and compliance of the autonomous driving algorithm based on each simulation test record to obtain the corresponding comprehensive score, and determining whether the intersection test corresponding to the current simulation test record has passed by using a preset scoring threshold, counting the number of passes / fails in the current simulation test task report, and rating the current evaluation algorithm based on the comprehensive score of all intersections and the total number of passes / fails according to a preset rating rule.

[0114] The rating rules here can be customized based on the actual needs of the application. For example, the comprehensive scores of all intersections can be summed and the total score can be normalized. At the same time, the pass rate can be calculated based on the total number of intersections that pass / fail. Then, the normalized total score and the pass rate can be weighted and summed to obtain a mixed total score. Finally, the level corresponding to the mixed total score can be confirmed based on different pre-set level segments.

[0115] It should also be noted that the specific steps of the embodiment of the present invention for comprehensively evaluating the safety, efficiency, comfort, and compliance of the autonomous driving algorithm based on each simulation test record include: taking each simulation test record as the corresponding current record; taking the intersection test scenario, test recording scenario, and vehicle object index of the current record as the corresponding current test scenario, current recording scenario, and current vehicle index; taking the scenario corresponding to the object index in the current test scenario, the current recording scenario, and the current vehicle index, and the recorded object as the corresponding vehicle tag object and vehicle test object; performing a safety evaluation based on the vehicle test object and the current recording scenario to obtain the corresponding safety score; performing an efficiency evaluation based on the vehicle tag object and the vehicle test object to obtain the corresponding efficiency score; performing a comfort evaluation based on the vehicle test object to obtain the corresponding comfort score; performing a compliance evaluation based on the current recording scenario and the vehicle test object to obtain the corresponding compliance score; and calculating the corresponding comprehensive score based on the obtained safety, efficiency, comfort, and compliance scores according to the preset comprehensive scoring rules. The comprehensive scoring rules here can be customized based on application needs. For example, the comprehensive score can be obtained by summing the scores of security, efficiency, comfort and compliance, or by weighted summing the scores of security, efficiency, comfort and compliance.

[0116] It should also be noted that the specific steps of obtaining the corresponding safety score based on the self-vehicle test object and the current recording scene in the embodiments of the present invention include: identifying whether there are trajectory points / time points in the self-vehicle trajectory with a collision time (TTC) of less than a set threshold in the self-vehicle test object and other scene objects around the self-vehicle, identifying whether the self-vehicle actively collides with other vehicles, identifying whether the self-vehicle is passively collided with other vehicles, and setting a safety score based on the three types of identification results according to the preset safety scoring rules. Here, when identifying whether a vehicle actively collides with another vehicle or is passively collided with by another vehicle, the system first identifies whether the coordinate distance between the vehicle's trajectory and the trajectory of other objects at the same time is less than a set distance threshold. If the coordinate distance at a certain moment is less than the distance threshold, a collision is confirmed and the corresponding coordinate at that moment is set as the collision position. If the coordinate distance throughout the entire time period is greater than or equal to the distance threshold, a collision is confirmed not to have occurred. If the vehicle's trajectory does not collide with any other object's trajectory, the identification results for both actively colliding with another vehicle and passively being collided with by another vehicle are both "not occurred". If it is confirmed that the vehicle's trajectory has collided with the trajectory of another object, the system identifies the traffic behavior type of the vehicle and the object before the collision by identifying the coordinates, orientation angle, speed, and acceleration of the vehicle and the object at the collision position. Then, according to the pre-set collision liability confirmation rules, the system confirms the vehicle's active / passive liability based on the traffic behavior type of the two parties, i.e., it confirms whether the vehicle actively collides with another vehicle or is passively collided with by another vehicle. After confirming the above three types of identification results, a safety score is set based on the safety scoring rules. The safety scoring rules can be customized based on application requirements. For example, a 0 / 1 binary rule can be customized for the three types of identification results of the above three types of events (collision time with the vehicle in front is too short, active collision with another vehicle, and passive collision with another vehicle): 0 for the event occurring and 1 for not occurring. The three scores are then summed to obtain the safety score.

[0117] It should also be noted that the specific steps of obtaining the corresponding efficiency score based on the efficiency evaluation of the vehicle tag object and the vehicle test object in this embodiment of the invention include: comparing the object state trajectories of the vehicle tag object and the vehicle test object to identify whether the vehicle has veered off course and whether the vehicle has experienced excessively long temporary parking time, and setting efficiency scores based on the two types of identification results according to preset efficiency scoring rules. When identifying whether the vehicle has veered off course, firstly, two coordinates at the same moment on the two object state trajectories of the vehicle tag object and the vehicle test object are combined into a coordinate pair, then the straight-line distance between each coordinate pair is calculated, then the total number of straight-line distances exceeding a set distance threshold is counted, and then it is identified whether the ratio of the total number of counted points to the total number of coordinate pairs exceeds a preset ratio threshold. If it exceeds the threshold, it is confirmed that a veered off course problem has occurred; otherwise, it is confirmed that no veered off course problem has occurred. When identifying whether a vehicle has experienced excessively long temporary parking time, the coordinates of the vehicle's trajectory where the speed is below a preset minimum speed threshold are recorded as parking positions. The trajectory segments corresponding to two or more consecutive parking positions are recorded as parking segments. The total number of parking segments is then identified. If the total number of parking segments is 0, no excessively long temporary parking time has occurred. If the total number of parking segments is greater than 0, the duration of each parking segment is further identified to see if it exceeds a preset parking duration threshold. If the duration of any parking segment exceeds the parking duration threshold, an excessively long temporary parking time has occurred. If the duration of all parking segments does not exceed the parking duration threshold, no excessively long temporary parking time has occurred. After confirming the above two types of identification results, an efficiency score is set based on the efficiency scoring rules. The efficiency scoring rules can be customized based on application requirements. For example, a 0 / 1 binary rule can be customized for the two types of identification results of the above two types of events (deviation and excessive temporary parking time): 0 for the event occurring and 1 for the event not occurring. The two scores are then summed to obtain the safety score.

[0118] It should also be noted that the specific steps of obtaining the corresponding comfort score based on the self-driving test object in the embodiments of the present invention include: identifying whether the self-driving vehicle has excessive acceleration, identifying whether the self-driving vehicle has excessive lateral speed, and identifying whether the self-driving vehicle has sudden braking behavior based on the object state trajectory of the self-driving test object, and setting a comfort score based on the three types of identification results according to the preset comfort scoring rules. When identifying whether a vehicle has experienced excessive acceleration, a pre-defined correspondence rule between speed ranges and acceleration rate of change ranges is used. This rule assigns an acceleration rate of change range to each speed range. Then, based on this rule, the acceleration rate of change range corresponding to the speed at each moment on the vehicle's trajectory is identified. Next, the vehicle's acceleration rate of change at each moment on the trajectory is calculated, and it is determined whether each acceleration rate of change meets the corresponding acceleration rate of change range. If the vehicle's acceleration rate of change at any moment does not meet the corresponding acceleration rate of change range, an excessive acceleration problem is confirmed; if the vehicle's acceleration rate of change at all moments meets the corresponding acceleration rate of change range, an excessive acceleration problem is confirmed. When identifying whether a vehicle has excessive lateral speed, the lateral speed is calculated based on the vehicle's speed, orientation angle, and the road travel direction corresponding to the vehicle's current coordinates on the high-precision map at each moment on the vehicle's trajectory. It then checks whether the lateral speed at each moment exceeds a preset lateral speed threshold. If any vehicle's lateral speed exceeds the threshold, an excessive lateral speed problem is confirmed; if none of the vehicle's lateral speeds exceed the threshold, no excessive lateral speed problem has occurred. Similarly, when identifying whether a vehicle has engaged in emergency braking, the acceleration at each moment on the vehicle's trajectory is checked against a preset emergency braking acceleration threshold. If any acceleration is below the threshold, an emergency braking action is confirmed; if all accelerations are within the threshold, no emergency braking action has occurred. Here, the emergency braking acceleration threshold is a threshold less than 0. After confirming the above three types of identification results, a comfort score is set based on the comfort scoring rules. The comfort scoring rules can be customized based on application requirements. For example, a 0 / 1 binary rule can be customized for the three types of identification results of the above three types of events (excessive acceleration, excessive lateral speed, and emergency braking): 0 for the event occurring and 1 for not occurring. The three scores are then summed to obtain the comfort score.

[0119] It should also be noted that the specific steps of obtaining the corresponding compliance score based on the current recording scenario and the self-driving test object in this embodiment of the invention include: identifying whether the self-driving vehicle has run a red light, whether it has left the crossing lane, and whether it has exceeded the speed limit, based on the object state trajectory of the self-driving test object and the high-precision map of the intersection in the current recording scenario; and setting a compliance score based on the three types of identification results according to the preset compliance scoring rules. When identifying whether the self-driving vehicle has run a red light, the red light area at each moment is given on the high-precision map of the intersection based on the object state trajectory of each traffic light object in the current recording scenario. It is confirmed whether the coordinates of the self-driving test object's object state trajectory at least once enter a red light area. If it is confirmed that at least one coordinate at a moment enters a red light area, it is confirmed that a red light has been run; if it is confirmed that no coordinates at any moment enter any red light area, it is confirmed that no red light has been run. When identifying whether a vehicle has left its lane, the system first identifies prohibited areas on the intersection's high-precision map, such as areas outside the road edge, areas outside double solid lane lines, and no-stopping zones. Then, it checks whether the vehicle's trajectory at least once enters a prohibited area. If at least one coordinate enters a prohibited area, the vehicle is confirmed to have left the lane; otherwise, it is confirmed that it has not left the lane. Similarly, when identifying whether a vehicle is speeding, the system first checks the speed limit thresholds on the vehicle's trajectory at each moment using the intersection's high-precision map. Then, it identifies whether the vehicle's speed at any moment exceeds the corresponding speed limit threshold. If the speed at any moment exceeds the corresponding speed limit threshold, speeding is confirmed; otherwise, speeding is confirmed. After confirming the above three types of identification results, a compliance score is set based on the compliance scoring rules. The compliance scoring rules can be customized based on application requirements. For example, a 0 / 1 binary rule can be customized for the three types of identification results of the above three events (running a red light, leaving the lane, and overtaking): 0 for the event occurring and 1 for not occurring. The three scores are then summed to obtain the compliance score.

[0120] Figure 5 This is a module structure diagram of a simulation test task processing device provided in Embodiment 2 of the present invention. This device can be a terminal device or server implementing the aforementioned method embodiments, or it can be a device that enables the aforementioned terminal device or server to implement the aforementioned method embodiments. For example, the device can be a device or chip system of the aforementioned terminal device or server. Figure 5As shown, the simulation test task processing device provided in Embodiment 2 of the present invention includes: a test task receiving module 201, an intersection scene preparation module 202, a simulation test module 203, and a task report saving module 204.

[0121] Test task receiving module 201 is used to receive test road R test Test period T test and test modes; and use the autonomous driving algorithm interface under test as the corresponding vehicle algorithm interface; test road R test This is a real road network; the test period is T. test The timing period is 24 hours; the test modes include the first and second modes.

[0122] The intersection scene preparation module 202 is used to identify the test mode; if the test mode is the first mode, then based on the test road R... test The real road network data will be used for the test period T test The road is divided into multiple consecutive intersection time segments, and test simulation scenarios are created for each intersection in its corresponding intersection time segment based on real road network data to obtain the corresponding intersection test scenarios; if the test mode is the second mode, then according to the test road R test Real road network data for each intersection on the current road during the test period T test The corresponding intersection test scenario is obtained by creating a test simulation scenario during the period of maximum pressure within the time frame.

[0123] The simulation test module 203 is used to take each intersection test scenario as the corresponding current intersection test scenario; and select a car object from the current intersection test scenario as the corresponding self-vehicle object based on the preset self-vehicle selection rules; and import the current intersection test scenario into the preset scene simulator for initialization; and connect the self-vehicle algorithm interface to the scene simulator to take over the driving control of the self-vehicle object; and the scene simulator performs a simulation test according to the loaded current intersection test scenario and the connected self-vehicle algorithm interface and outputs the corresponding test recording scenario; the self-vehicle selection rules include at least random, long-range and high-efficiency.

[0124] The task report saving module 204 is used to use the object index of the vehicle object in each test recording scenario as the corresponding vehicle object index; and to form a corresponding simulation test record by combining each test recording scenario with its corresponding vehicle object index and the intersection test scenario; and to form a corresponding simulation test task report by combining all the obtained simulation test records and saving it.

[0125] It should be noted that the simulation test task processing device provided in Embodiment 2 of the present invention may further include an algorithm evaluation module; the algorithm evaluation module is used to, after obtaining the simulation test task report, comprehensively evaluate the safety, efficiency, comfort and compliance of the autonomous driving algorithm based on each simulation test record to obtain the corresponding comprehensive score, and determine whether the intersection test corresponding to the current simulation test record has passed through a preset scoring threshold, count the number of passes / fails in the current simulation test task report, and rate the current evaluation algorithm based on the comprehensive score of all intersections and the total number of passes / fails according to the preset rating rules.

[0126] The simulation test task processing device provided in this embodiment of the invention can execute the method steps in the above method embodiment. Its implementation principle and technical effect are similar, and will not be repeated here.

[0127] It should be noted that the division of the various modules in the above device is merely a logical functional division. In actual implementation, they can be fully or partially integrated into a single physical entity, or they can be physically separated. Furthermore, these modules can be implemented entirely in software via processing element calls; they can be fully implemented in hardware; or some modules can be implemented by processing element calls to software, while others are implemented in hardware. For example, the test task receiving module can be a separate processing element, or it can be integrated into a chip in the above device. Alternatively, it can be stored as program code in the memory of the above device, and called and executed by a processing element of the device. The implementation of other modules is similar. Moreover, these modules can be fully or partially integrated together, or they can be implemented independently. The processing element described here can be an integrated circuit with signal processing capabilities. In the implementation process, each step of the above method or each of the above modules can be completed through integrated logic circuits in the hardware of the processor element or through software instructions.

[0128] For example, these modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more Digital Signal Processors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs). As another example, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a System-on-a-Chip (SOC).

[0129] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. This computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the foregoing method embodiments are generated. The computer described above can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The aforementioned computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the aforementioned computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, Bluetooth, microwave, etc.) means. The aforementioned computer-readable storage medium can be any available medium that a computer can access, or a data storage device such as a server or data center that integrates one or more available media. The aforementioned available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state disks (SSDs)).

[0130] Figure 6 This is a schematic diagram of an electronic device provided in Embodiment 3 of the present invention. This electronic device can be a terminal device or server implementing the methods of the aforementioned embodiments, or it can be a terminal device or server connected to the aforementioned terminal device or server implementing the methods of the aforementioned embodiments. Figure 6As shown, the electronic device may include: a processor 301 (e.g., CPU), a memory 302, and a transceiver 303; the transceiver 303 is coupled to the processor 301, and the processor 301 controls the transmission and reception operations of the transceiver 303. The memory 302 may store various instructions for performing various processing functions and implementing the processing steps described in the foregoing embodiments. Preferably, the electronic device involved in the embodiments of the present invention further includes: a power supply 304, a system bus 305, and a communication port 306. The system bus 305 is used to realize communication connections between components. The communication port 306 is used for communication between the electronic device and other peripherals.

[0131] exist Figure 6 The system bus 305 mentioned can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This system bus can be divided into address bus, data bus, control bus, etc. For ease of representation, it is represented by only one thick line in the figure, but this does not indicate that there is only one bus or one type of bus. The communication interface is used to enable communication between the database access device and other devices (e.g., clients, read-write libraries, and read-only libraries). Memory may include Random Access Memory (RAM) and may also include non-volatile memory, such as at least one disk storage device.

[0132] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), graphics processing units (GPUs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0133] It should be noted that the embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when run on a computer, cause the computer to perform the methods and processes provided in the above embodiments.

[0134] This invention provides a method, apparatus, electronic device, and computer-readable storage medium for processing simulation test tasks. As described above, this invention provides two methods for creating realistic intersection scenarios to enhance scenario complexity and diversity: continuous scenario creation for all intersections on any test road based on time continuity, and stress scenario creation for all intersections on any test road based on traffic stress test characteristics. After scenario creation, simulation-testing is achieved by connecting the autonomous vehicle algorithm interface to the scenario simulator. Three types of autonomous vehicle selection rules (random, long-range, and efficient) are provided for each scenario test to improve the comprehensiveness of the test. The scenario simulator performs simulation testing of the autonomous driving algorithm based on the currently loaded intersection test scenario and through the connected autonomous vehicle algorithm interface, and records the real-time test scenario during the test. This invention enhances the realism, complexity, and diversity of test scenarios, expands the types and range of test vehicles, and effectively improves the complexity and comprehensiveness of algorithm testing.

[0135] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented in hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0136] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for processing simulation test tasks, characterized in that, The method includes: Receive test road R test Test period T test and test mode; and use the autonomous driving algorithm interface to be tested as the corresponding vehicle algorithm interface; the test road R test The test period T is a real traffic network road segment. test The timing period is 24 hours; the test mode includes a first mode and a second mode; The test mode is identified; if the test mode is the first mode, then according to the test road R... test The real road network data will be used for the test period T test The road is divided into multiple consecutive intersection time segments, and test simulation scenarios are created for each intersection in its corresponding intersection time segment based on real road network data to obtain corresponding intersection test scenarios; if the test mode is the second mode, then according to the test road R test Real road network data for each intersection on the current road during the test period T test The corresponding intersection test scenario is obtained by creating a test simulation scenario during the period of maximum pressure within the time frame. Each intersection test scenario is used as the corresponding current intersection test scenario; and a car object is selected from the current intersection test scenario as the corresponding autonomous vehicle object based on a preset autonomous vehicle selection rule; the current intersection test scenario is imported into a preset scene simulator for initialization; and the autonomous vehicle algorithm interface is connected to the scene simulator to take over the driving control of the autonomous vehicle object; and the scene simulator performs a simulation test according to the loaded current intersection test scenario and the connected autonomous vehicle algorithm interface and outputs the corresponding test recording scenario; the autonomous vehicle selection rule includes at least random, long-range, and high-efficiency; The object index of the vehicle object in each of the test recording scenarios is used as the corresponding vehicle object index; and each of the test recording scenarios, its corresponding vehicle object index, and the intersection test scenario are combined to form a corresponding simulation test record; and all the obtained simulation test records are combined to form a corresponding simulation test task report and saved. Among them, the test road R test Including multiple intersections C i and multiple road sections j 1 ≤ intersection index i ≤ N, 1 ≤ road segment index j ≤ N-1, where N is the total number of intersections; the test road R test The starting and ending positions are at intersection C. i=1 C i=N The road segment r j Intersection C i=j C i=j+1 The connecting road sections between; the test period T test The start and end times are denoted as t. start t end The test period T test Test duration L test =t end -t start ; Each of the aforementioned intersections C i The corresponding intersection time segment is denoted as TA. i TA i ∈T test The two adjacent intersection time slots TA before and after. i TA i+1 In this process, the end time of the preceding time period is aligned with the start time of the following time period; the intersection segmentation time TA i The start and end times are denoted as ta s,i ,ta e,i ,ta s,i=1 =t start ,ta s,i+1 =ta e,i ,ta e,i=N =t end ; Each of the aforementioned intersections C i The corresponding period of maximum pressure is denoted as TB. i TB i ∈T test The maximum pressure period TB i The start and end times are denoted as tb. s,i tb e,i ; Each of the aforementioned intersections C i The corresponding intersection test scenario and the corresponding test recording scenario are denoted as the corresponding , The intersection test scenario The data content includes at least a high-precision map of the intersection. and scene object set The high-precision map of the intersection The map elements include at least the traffic signs / markers / markings elements, road edge line elements, danger zone elements, obstacle elements, and traffic light elements for all roads at the current intersection; the scene object set Includes multiple scene objects 1 ≤ object index o ≤ N D,i N D,i The total number of objects in the test scenario at the i-th intersection; According to the test road R test The real road network data will be used for the test period T test The time segment is divided into multiple consecutive intersection segments, specifically including: Based on the test road R test The real road network data is used to analyze each of the aforementioned road segments r within a preset specified time period. j During the test period T test The average vehicle speed within the area is statistically analyzed to obtain the corresponding average vehicle speed. ; and each of the aforementioned road segments r j The length of the straight road is denoted as the corresponding length d. j ; and based on each of the aforementioned lengths d j and the corresponding average vehicle speed Calculate the corresponding average travel time ; and for N-1 of the average travel times Δl j The total duration is calculated by summing the total duration. sum The specified time period includes at least a specified single day or multiple consecutive days, a specified single week or multiple consecutive weeks, a specified single month or multiple consecutive months, a specified single quarter or multiple consecutive quarters, or a specified single year or multiple consecutive years. And the total duration l sum With the test duration L test Perform identification; if the total duration is l sum Less than the test duration L test Then set the corresponding end intersection segment duration △l end =L test -l sum If the total duration is l sum Greater than or equal to the test duration L test Then, any one of the average travel times Δl can be selected. j As the corresponding incremental duration △l add and the test period T test End time t end The test duration L is postponed to make the latest test duration L test Satisfy L test =l sum +△l add And set the corresponding segmentation time Δl at the tail intersection. end =△l add ; And based on the obtained N-1 average travel times Δl j The segmentation time Δl at the tail intersection end and the test period T test start time t start and the end time t end For the N intersection time segments TA i The start time ta s,i and the end time ta e,i Configure settings: , , , , , ; According to the test road R test Real road network data for each intersection on the current road during the test period T test The test simulation scenario during the period of maximum pressure within the time frame is used to create the corresponding intersection test scenario, specifically including: Each of the aforementioned intersections C i As the corresponding current intersection; and based on the test road R test The real road network data is used to test the current intersection during the specified test period T each day within a preset time period. test The maximum number of traffic participants within the area is used to calculate the corresponding second total; and the largest second total is denoted as the total y. max,i ; and the total number y max,i The corresponding daily road network data is related to the current intersection and the test period T. test The corresponding regional time period road network data is used as the corresponding current intersection data; and the total number y is used as the data. max,i During the test period T test The corresponding time in the table is taken as the corresponding maximum pressure period TB. i start time tb s,i and the test period T test The end time t end The current maximum pressure period TB i End time tb e,i ; and the current intersection data that meets the current maximum pressure period TB i The road network data is used as the corresponding intersection pressure period data; and each of the aforementioned intersections C i The corresponding high-precision map serves as the corresponding high-precision map of the intersection. ; and each of the aforementioned intersections C i The corresponding intersection pressure time period data is used as the corresponding current pressure time period data; and the motion trajectory of each car, motorcycle, bicycle or pedestrian in the current pressure time period data is sampled according to a preset trajectory sampling frequency, and the corresponding scene object is constructed based on the sampling results. The coordinates and changing trajectories of each traffic light in the current pressure period data are sampled according to the trajectory sampling frequency, and the corresponding scene object is constructed based on the sampling results. ; and from all the scene objects obtained Form the corresponding scene object set ; and the obtained high-precision map of the intersection and the set of scene objects The corresponding intersection test scenario is formed. .

2. The method for processing simulation test tasks according to claim 1, characterized in that, The input data for the autonomous vehicle algorithm interface includes simulation timestamps, traffic participant datasets, traffic light datasets, autonomous vehicle status data, autonomous vehicle target position, and high-precision map data. The output is vehicle driving control commands. The traffic participant dataset consists of one or more traffic participant data sets, which include participant type, participant coordinates, participant speed, participant orientation angle, and participant acceleration. The traffic light dataset consists of one or more traffic light data sets, which include traffic light coordinates and lighting status, including red, green, yellow, and flashing yellow. The autonomous vehicle status data includes autonomous vehicle coordinates, autonomous vehicle orientation angle, autonomous vehicle speed, and autonomous vehicle acceleration. The autonomous vehicle target position is a target position coordinate. The command parameters of the vehicle driving control commands include at least steering wheel angle, accelerator pedal opening, brake pedal opening, and gear position. The test recording scenario With the intersection test scenario One-to-one correspondence; the intersection test scenario The test recording scenario The data encapsulation format all conforms to the OpenSCENARIO standard scene file data encapsulation format; The scene object Including object types Object state trajectory { }, where t is a timestamp; the object type At least including cars, motorcycles, bicycles, pedestrians, and traffic lights; the state trajectories of all said objects { The trajectory sampling frequency 1 / Δt remains consistent; the object type When representing cars, motorcycles, bicycles, and pedestrians, the object state is... It includes at least coordinates, orientation angle, velocity, and acceleration; the object type When it is a traffic light, the object state This includes the coordinates of the traffic lights and the status of the traffic lights, wherein the status of the traffic lights includes at least red light, green light, yellow light, and flashing yellow light; The test recording scenario The data content includes at least a high-precision map of the intersection. and recording object set The high-precision map of the intersection The corresponding high-precision map of the intersection Maintain consistency; the set of recorded objects Including N D,i Recording objects The recording object With the scene object One-to-one correspondence; the recorded object Including object types and object state trajectory { }; the object type With the corresponding object type Maintain consistency; The object's state trajectory { } and the corresponding object state trajectory { Timestamp alignment; the object type When the object is a car, motorcycle, bicycle, or pedestrian, the object state is as follows: It includes at least coordinates, orientation angle, velocity, and acceleration; the object type When it is a traffic light, the object state This includes the traffic light coordinates and the traffic light status, wherein the traffic light status includes at least red light, green light, yellow light, and flashing yellow light; The scene simulator is a type of scene simulation tool that conforms to the OpenSCENARIO standard.

3. The method for processing simulation test tasks according to claim 2, characterized in that, The process of creating corresponding intersection test scenarios based on real road network data for each intersection in its corresponding intersection time segment specifically includes: Based on the test road R test The real road network data is used to measure the current road during the pre-set specified time period T each day. test The total number of traffic flows is counted to obtain the corresponding first total, and the single-day road network data corresponding to the largest first total is used as the road network data for that day. And the road network data of the day and each of the intersections C i The corresponding regional road network data is used as the corresponding intersection area data; and the intersection area data is divided into the corresponding intersection time slots (TA). i The time-period road network data is used as the corresponding intersection time-period data; and each of the aforementioned intersections C i The corresponding high-precision map serves as the corresponding high-precision map of the intersection. ; and each of the aforementioned intersections C i The corresponding intersection time period data is used as the corresponding current intersection time period data; and the motion trajectory of each car, motorcycle, bicycle or pedestrian in the current intersection time period data is sampled according to a preset trajectory sampling frequency, and the corresponding scene object is constructed based on the sampling results. The system samples the coordinates and changing trajectories of each traffic light in the current intersection time period data according to the trajectory sampling frequency, and constructs the corresponding scene object based on the sampling results. ; and from all the scene objects obtained Form the corresponding scene object set ; and the obtained high-precision map of the intersection and the set of scene objects The corresponding intersection test scenario is formed. .

4. The method for processing simulation test tasks according to claim 2, characterized in that, The method of selecting a car object from the current intersection test scenario based on a preset vehicle selection rule as the corresponding vehicle object specifically includes: The scene object set of the current intersection test scenario. The object type described in The scene object for the car The objects are clustered into a corresponding set of car objects; the vehicle selection rule is identified; if the vehicle selection rule is random, a scene object is randomly selected from the set of car objects. As the corresponding vehicle object; if the vehicle selection rule is long-range, then the object state trajectory in the vehicle object set { The longest of the aforementioned scene objects As the corresponding vehicle object; if the vehicle selection rule is efficient, then the object state trajectory in the vehicle object set { The scene object with the fastest average speed As the corresponding vehicle object.

5. The method for processing simulation test tasks according to claim 2, characterized in that, The step of importing the current intersection test scenario into a preset scenario simulator for initialization specifically includes: The scene simulator uses the high-precision map of the intersection in the currently imported test scene. Construct a two-dimensional or three-dimensional simulation world; and based on the scene object set of the current intersection test scenario. Each of the aforementioned scene objects In the current simulation world, a one-to-one simulation object is constructed; and the simulation time period of the current simulation world is configured based on the time span of the current intersection test scenario; and the recording sampling frequency and single-step simulation frequency of the current simulation world are configured based on the trajectory sampling frequency of the current intersection test scenario, wherein the recording sampling frequency is consistent with the current trajectory sampling frequency, and the single-step simulation frequency is an integer multiple of the current trajectory sampling frequency; and simulation objects of the type of car, motorcycle, bicycle, or pedestrian in the current simulation world are regarded as participant objects, and corresponding traffic participant behavior models, motion control models, and dynamic models are assigned to each participant object; and simulation objects of the type of traffic light are regarded as traffic light objects, and corresponding traffic light control models are assigned to each traffic light object; and the simulation is based on the object state trajectory corresponding to each simulation object { The first object state in} The initial simulation state of the current simulation object is set, and the second to last object states of each simulation object are set. The multi-stage simulation objective serves as the current simulation object; The traffic participant behavior model is used to determine the traffic behavior type of the current participant at a future simulation time based on the current simulation goal and current historical trajectory of each participant at the current simulation time, the historical trajectories of other participants at the current simulation time, and the historical trajectories of surrounding traffic light objects at the current simulation time. The traffic behavior types include at least decelerating straight ahead, accelerating straight ahead, driving straight ahead at a constant speed, changing lanes to the left, changing lanes to the right, overtaking, making a U-turn, yielding to the left, yielding to the right, sudden braking, and stopping. The motion control model is used to predict the motion control parameters of each participant at a future simulation time based on the current simulation goal, current traffic behavior type, and current historical trajectory at the current simulation moment; the motion control parameters include at least steering force parameters, longitudinal acceleration force parameters, and longitudinal deceleration force parameters; The dynamic model is used to predict the simulation state of the current participant at future simulation moments based on the current simulation target, current simulation state, and current motion control parameters of each participant at the current simulation moment; the simulation state includes at least coordinates, orientation angle, velocity, and acceleration; The traffic light control model is used to predict the traffic light state of the current traffic light object at future simulation times based on the current simulation target and current historical trajectory of each traffic light object at the current simulation time.

6. The method for processing simulation test tasks according to claim 5, characterized in that, The step of connecting the autonomous vehicle algorithm interface to the scene simulator to take over the driving control of the autonomous vehicle specifically includes: The simulation object corresponding to the autonomous vehicle in the current simulation world of the scene simulator is taken as the corresponding current takeover object; the traffic participant behavior model and motion control model assigned to the current takeover object by the scene simulator are canceled; the input end of the dynamics model assigned to the current takeover object by the scene simulator is connected to the output end of the autonomous vehicle algorithm interface; and a corresponding simulation data output interface is set in the scene simulator and connected to the input end of the autonomous vehicle algorithm interface. The simulation data output interface is used to, at each simulation moment in the simulation process, take the real-time simulation state of the currently taking over object in the current simulation world as the corresponding vehicle state data; take the real-time simulation state of each of the other participating objects around the currently taking over object as the corresponding traffic participant data, and form a corresponding traffic participant dataset from all the obtained traffic participant data; take the coordinates and real-time traffic light state of each traffic light object in the current simulation world as the corresponding traffic light coordinates and lighting state to form a corresponding traffic light data, and form a corresponding traffic light dataset from all the obtained traffic light data; take the coordinate position of the most recent next-stage simulation target in the multi-stage simulation target corresponding to the currently taking over object as the corresponding vehicle target position; and take the high-precision map of the intersection corresponding to the current simulation world. The high-precision map data is used as the corresponding data; the current simulation time is used as the corresponding simulation timestamp; and the obtained simulation timestamp, traffic participant dataset, traffic light dataset, vehicle status data, vehicle target location, and high-precision map data are used as the output data for this simulation and output to the vehicle algorithm interface.

7. The method for processing simulation test tasks according to claim 5, characterized in that, The process of the scene simulator performing a simulation test based on the loaded current intersection test scenario and the accessed autonomous vehicle algorithm interface, and outputting the corresponding test recording scenario, specifically includes: During the simulation process, the scene simulator simulates the motion trajectory of the current participant object using the traffic participant behavior model, motion control model, and dynamics model of each participant object; it also simulates the changing trajectory of the current traffic light object using the traffic light control model of each traffic light object; the simulation data output interface, vehicle algorithm interface, and dynamics model of the vehicle object simulate the motion trajectory of the vehicle; and based on the recording sampling frequency, it collects data on the real-time simulation state of each simulation object at each recording sampling point to generate the corresponding object state. At the end of the simulation, the high-precision map of the intersection in the current intersection test scenario will be generated. As the corresponding high-precision map of the intersection and the object type corresponding to each of the simulation objects. As a corresponding object type And the state of all objects corresponding to each of the simulation objects. Form a corresponding object state trajectory { }, and by the object type corresponding to each of the simulation objects. and the object state trajectory { } to form a corresponding recording object And by all the recorded objects The corresponding set of recording objects is formed. And from the high-precision map of the intersection and the set of recorded objects The corresponding test recording scenario is formed. And output it.

8. An apparatus for performing a processing method for a simulation test task according to any one of claims 1-7, characterized in that, The device includes: a test task receiving module, an intersection scene preparation module, a simulation test module, and a task report saving module; The test task receiving module is used to receive the test road R. test Test period T test and test mode; and use the autonomous driving algorithm interface to be tested as the corresponding vehicle algorithm interface; the test road R test The test period T is a real traffic network road segment. test The timing period is 24 hours; the test mode includes a first mode and a second mode; The intersection scene preparation module is used to identify the test mode; if the test mode is the first mode, then according to the test road R test The real road network data will be used for the test period T test The road is divided into multiple consecutive intersection time segments, and test simulation scenarios are created for each intersection in its corresponding intersection time segment based on real road network data to obtain corresponding intersection test scenarios; if the test mode is the second mode, then according to the test road R test Real road network data for each intersection on the current road during the test period T test The corresponding intersection test scenario is obtained by creating a test simulation scenario during the period of maximum pressure within the time frame. The simulation testing module is used to take each intersection test scenario as the corresponding current intersection test scenario; and select a car object from the current intersection test scenario as the corresponding autonomous vehicle object based on a preset autonomous vehicle selection rule; import the current intersection test scenario into a preset scene simulator for initialization; and connect the autonomous vehicle algorithm interface to the scene simulator to take over the driving control of the autonomous vehicle object; and the scene simulator performs a simulation test according to the loaded current intersection test scenario and the connected autonomous vehicle algorithm interface and outputs the corresponding test recording scenario; the autonomous vehicle selection rule includes at least random, long-range, and high-efficiency; The task report saving module is used to use the object index of the vehicle object in each of the test recording scenarios as the corresponding vehicle object index; and to form a corresponding simulation test record by combining each of the test recording scenarios, its corresponding vehicle object index, and the intersection test scenario; and to form a corresponding simulation test task report by combining all the obtained simulation test records and saving it.

9. An electronic device, characterized in that, include: Memory, processor, and transceiver; The processor is configured to be coupled to the memory, read and execute instructions in the memory to implement the method according to any one of claims 1-7; The transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a computer, cause the computer to perform the method described in any one of claims 1-7.

Citation Information

Patent Citations

  • Simulation traffic scene file generation method and device

    CN112567374A

  • Interactive intelligent vehicle traffic simulation test method and system based on cloud service

    CN114004113A