Acceleration test method for continuous conflict scene of closed site of multiple traffic participants

Abstract

The invention relates to an accelerated test method for a continuous conflict scene of a multi-traffic-participant closed site. The accelerated test method comprises the following steps: generating a traffic participant global path based on a tested vehicle global path; based on the global paths of the tested vehicle and the traffic participants, interaction areas of the tested vehicle and the traffic participants are constructed, and in each interaction area, test tasks and priorities are distributed to the traffic participants; constructing an SL graph and an ST graph for each traffic participant based on the known test task and the priority, and generating an optimal local trajectory of the traffic participant based on the SL graph and the ST graph; and after the optimal local trajectory executed by each traffic participant is obtained, judging whether the test is ended or not according to the execution conditions of all the test tasks and the collision condition with the tested vehicle. Compared with the prior art, local path distribution is carried out on each traffic participant based on collision avoidance among multiple traffic participants and an active test task of the tested vehicle, and the closed site continuous conflict scene acceleration test is realized on the premise of ensuring the driving safety of the traffic participants.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous vehicle testing, and in particular to an acceleration testing method for a continuous conflict scenario of multiple traffic participants in a closed field. Background Art

[0002] Closed-field testing is an important safety verification link before the implementation of autonomous vehicles. With the rapid development of autonomous driving technology, low-level and low-intelligence closed-field testing methods are difficult to evaluate high-level and high-intelligence autonomous vehicles, and there are many problems with existing closed-field testing methods, specifically: the testing equipment (crash test dummies, dummy vehicles) used to simulate traffic participants has low intelligence and there is no coordination during the testing process, resulting in poor interactivity of the autonomous driving test system and difficulty in reflecting the intelligence level of autonomous vehicles; for testing methods for specific autonomous driving functions, there are problems such as scattered test conditions, simple scenario conditions, and fragmented function testing, resulting in most test scenarios being oriented towards a single condition, and the site mainly conducts fragmented and piecemeal testing, making it difficult to reflect the response ability of autonomous vehicles in complex traffic environments; for testing methods based on random traffic flow models, there are problems such as high costs, vague specific requirements for scenarios, and random and inefficient conflicts, resulting in long testing times for autonomous vehicles, low probability of encountering marginal scenarios, and difficulty in reflecting the intelligence level of autonomous vehicles in marginal scenarios. Taking the Chinese patent application CN118092393A as an example, although it realizes the simulation of traffic participant clusters and continuous interaction between test vehicles through an autonomous vehicle cloud control test system established by 5G communication and cloud computing technologies, and completes tests with high integrity, close to real traffic environments, covering high-risk, low-probability, and difficult-to-replicate scenarios, it still has the following disadvantages: 1) The motion control of the simulated traffic participants completely depends on preset instructions and cannot achieve active coordination among multiple traffic participants; 2) Taking roundabout testing as an example, the test scenarios are combinations of discrete scenarios such as roundabout entry, roundabout following, and roundabout exit, and the test scenarios are fragmented, making it difficult to reproduce the complex conflicts of continuous interaction in real traffic.

[0003] Therefore, it is a technical problem to be solved to provide an autonomous vehicle testing method that can achieve testing based on actively coordinated multiple traffic participants and continuous complex scenarios. Summary of the Invention

[0004] The purpose of the present invention is to provide an acceleration testing method for a continuous conflict scenario of multiple traffic participants in a closed field. A continuous complex test scenario is constructed through the global path of the vehicle under test, and in this continuous complex test scenario, local trajectory optimization is performed for each traffic participant with the goal of collision avoidance between multiple traffic participants and the active test task of the vehicle under test, so as to achieve the acceleration testing of the vehicle under test in a continuous complex conflict scenario.

[0005] The object of the present invention can be achieved by the following technical solutions: The present invention provides an accelerated test method for continuous conflict scenarios of multiple traffic participants in a closed area, and the method includes: Obtain the global path of the vehicle under test, construct a continuous conflict scenario based on the global path of the vehicle under test, and generate the global path of traffic participants by using the optimized A* algorithm based on the continuous conflict scenario; Construct an interaction area between the vehicle under test and traffic participants based on the global path of the vehicle under test and the global path of traffic participants, and construct an optimized model based on the auction algorithm. For each of the interaction areas, use the optimized model to assign test tasks and priorities of traffic participants to traffic participants; For each traffic participant, perform the following steps to obtain the optimal local trajectory: Obtain the vehicle states of the vehicle under test and traffic participants with priorities lower than the target traffic participant during the execution of all assigned test tasks by the target traffic participant, perform obstacle projection based on the vehicle states of traffic participants with priorities lower than the target traffic participant, perform task projection based on the vehicle state of the vehicle under test, and construct the SL map and ST map of the target traffic participant; Obtain the road information in the test scenario, construct the kinematic model and its constraints of the target traffic participant based on the road information, and obtain the real-time state of the target traffic participant based on the kinematic model and its constraints; the road information includes road curvature; Construct a multi-objective optimization function for local trajectory allocation, sample trajectory points of the target traffic participant in the SL map and ST map based on the real-time state, and generate trajectory clusters by combining the trajectory points. Each trajectory cluster includes multiple trajectories. Calculate the multi-objective optimization function for local trajectory allocation of each trajectory, and select the trajectory corresponding to the optimal value as the optimal local trajectory; After obtaining the execution of all test tasks and the collision situation between the vehicle under test and traffic participants after each traffic participant executes the optimal local trajectory, if all test tasks are executed or the vehicle under test collides with traffic participants, the test ends; otherwise, update the interaction area based on the optimal local trajectory and re-allocate test tasks.

[0006] As a preferred technical solution, the method for generating the global path of traffic participants is: Obtain key waypoints based on the global path of the vehicle under test, select path points by using the optimized A* algorithm based on the key waypoints, and generate the global path of traffic participants based on the path points; Among them, the method for selecting path points is to calculate the comprehensive priority of candidate path points, and select path points based on the comprehensive priority. Its expression is: , , , , Among them, represents the distance from the to-be-selected path point n to the starting point; represents the estimated distance from the to-be-selected path point n to the end point; represents the sum of the distances from the to-be-selected path point n to all the key path points; represents the comprehensive priority of the i-th path point; represents the position of the to-be-selected path point n; represents the position of the end point; represents the position of the j-th key path point; represents the number of key path points.

[0007] As a preferred technical solution, the method for allocating the test tasks of traffic participants and the priority when performing the described test tasks is as follows: Based on the global path of the vehicle under test and the global paths of traffic participants, obtain the interaction area between the vehicle under test and traffic participants, and define a set of test tasks for the interaction area; Based on the optimization model of the auction algorithm and the set of test tasks, perform the allocation of traffic participant test tasks. The expression of the optimization model is: , , Among them, represents the net benefit when traffic participant i performs test task j, obtained based on the distance of traffic participant i from the point corresponding to test task j; N represents the number of traffic participants; represents the number of test tasks; represents the test task allocation status of traffic participant i. If then it means that traffic participant i is assigned to perform test task j. If then it means that traffic participant i does not perform test task j; represents the set of traffic participants; represents the upper limit value of the number of traffic participants assigned to perform test tasks; Based on the global path of the vehicle under test, obtain the order in which the vehicle under test passes through the road topology structure, and determine the priority of performing test tasks based on the order.

[0008] As a preferred technical solution, the method for constructing the SL graph and ST graph of the target traffic participant includes: Convert the target traffic participant from the Cartesian coordinate system to the Frenet coordinate system. In the Frenet coordinate system, layer and decouple the trajectory of the target traffic participant into an SL graph and an ST graph; Traverse the trajectories of other traffic participants to obtain the potential interaction range, and filter out the traffic participants with a lower priority than the target traffic participant within the potential interaction range, and project them onto the SL graph and the ST graph as obstacles respectively; Calculate the task time domain when the target traffic participant executes each assigned test task based on the vehicle state of the vehicle under test, that is, the time when the vehicle under test reaches the interaction area corresponding to the test task; Obtain the physical sizes of the vehicle under test and the traffic participants, project the task time domain corresponding to each assigned test task onto the T-axis of the ST graph based on the physical sizes, and project the interaction area corresponding to the test task onto the S-axis of the SL graph and the ST graph to achieve task projection.

[0009] As a preferred technical solution, the projection method of the obstacle is as follows: Perform geometric dilation modeling on the obstacle to obtain a geometric dilation model, and its expression is: , , , , where The right front corner point of the obstacle; The left front corner point of the obstacle; Indicates the left rear corner point of the obstacle; Indicates the right rear corner point of the obstacle; Indicates the center coordinate point of the obstacle; The heading angle, that is, the angle between the traveling direction of the obstacle and the axis angle; Indicates half of the longitudinal length of the obstacle; Indicates half of the lateral length of the obstacle; Indicates the longitudinal dilation coefficient; Indicates the lateral dilation coefficient; Convert the coordinates of the geometric dilation model to the Frenet coordinate system, and project the geometric dilation model in the Frenet coordinate system onto the SL graph and the ST graph.

[0010] As a preferred technical solution, the projection method of the task time domain is as follows: , , where Indicates the time when the vehicle under test arrives at the interaction area, and , Indicates the distance from the vehicle under test to the interaction area, Indicates the speed of the vehicle under test; Indicates the projection range of the task time domain on the T axis of the ST diagram; Indicates half of the width of the target traffic participant.

[0011] As a preferred technical solution, the projection method of the interaction area is: , , Among them, Indicates the position where the traffic participant arrives at the interaction area; Indicates half of the width of the vehicle under test; Indicates the projection range of the interaction area projected onto the S axis.

[0012] As a preferred technical solution, the kinematic model and its constraints are: ， , , , , Among them, Indicates the derivative of the longitudinal distance of the traffic participant in the Frenet coordinate system; Indicates the speed of the traffic participant; Indicates the heading angle of the traffic participant; Indicates the curvature of the road in the Frenet coordinate system; Indicates the lateral distance traveled by the traffic participant in the Frenet coordinate system.

[0013] As a preferred technical solution, the local trajectory allocation multi-objective optimization function is: , , , , , ,

[0014] , , , , wherein, represents the allocated multi-objective cost of the i-th trajectory in the described trajectory cluster; represents the centripetal acceleration weight; represents the centripetal acceleration cost; represents the lateral driving cost; represents the longitudinal driving cost; represents the lateral offset weight; represents the lateral offset cost; represents the speed at the t-th moment; represents the road curvature; represents the total time required for the driving trajectory; represents the lateral offset weight; represents the lateral offset cost; represents the lateral comfort weight; represents the lateral comfort cost; represents the longitudinal position at the t-th moment; represents the lateral offset at the longitudinal position s; represents the lateral error boundary; represents the direction factor with the initial lateral offset; represents the second derivative of the lateral offset at the longitudinal position s; represents the longitudinal speed at the t-th moment; represents the first derivative of the lateral offset at the longitudinal position s; represents the longitudinal acceleration at the t-th moment; represents the weight to reach the end point of the trajectory; represents the cost to reach the end point of the trajectory; represents the weight to avoid collision; represents the cost to avoid collision; represents the smoothing weight; represents the smoothing cost; represents the total number of trajectory points; represents the sampling time of the j-th trajectory point; represents the deviation of the j-th trajectory point from the desired speed; represents the length of the i-th trajectory in the trajectory cluster; represents the Gaussian function of the distance cost between the j-th trajectory point and the obstacle projection; represents the longitudinal acceleration of the j-th trajectory point; represents the maximum value of the longitudinal acceleration of the j-th trajectory point; A Gaussian function representing the distance cost between the trajectory point j and the test task.

[0015] As a preferred technical solution, the method for obtaining the optimal local trajectory further includes: performing collision detection and task detection on the trajectory corresponding to the optimal value. If the requirements are met, the trajectory corresponding to the optimal value is the optimal local trajectory; otherwise, delete the trajectory corresponding to the optimal value from the trajectory cluster and re-select the optimal local trajectory. Among them, the collision detection is: traversing all obstacle projections in the SL graph and the ST graph, and determining whether the trajectory corresponding to the optimal value intersects with any of the obstacle projections. If so, it is determined that there is a collision; otherwise, there is no collision. The task detection is: traversing all test task projections in the ST graph, and determining whether the trajectory corresponding to the optimal value intersects with any of the test task projections. If so, it is determined that the corresponding traffic participant performs the test task; otherwise, it is determined that the corresponding traffic participant does not perform the test task.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1). The present invention allocates the global path of traffic participants through the global path of the vehicle under test, and divides the interaction area according to the global paths of the two. For each interaction area, an optimization model based on the auction algorithm is used for task allocation of traffic participants and priority division of traffic participants in the corresponding tasks, increasing the initiative of traffic participants in the interaction decision-making process. Through the priority competition mechanism, the order of multiple traffic participants performing the same test task is dynamically adjusted, reducing the absolute dependence on preset instructions, improving the flexibility of multiple traffic participants performing test tasks. 2). This application generates a continuous conflict scenario based on the global path of the vehicle under test. In this continuous conflict scenario, the global path of traffic participants is generated based on the key waypoints in the global path of the vehicle under test, and local trajectory allocation is performed in the Frenet coordinate system by combining the kinematic model and constraints. When screening the optimal local path, by integrating collision avoidance between multiple traffic participants and the active test task of the vehicle under test, quantitative characterization of safe interaction between vehicles and active test tasks is realized, enabling the method provided by the present invention to obtain the best test task execution strategy within a safe range, ensuring driving safety and accelerating the accelerated test of the closed continuous conflict scenario.

[0017] 3). The present invention separately solves the local optimal path for each traffic participant among multiple traffic participants participating in the test task, reducing the complexity of the optimization problem through distributed computing, lightening the calculation, and improving the calculation speed. Description of the Drawings

[0018] Figure 1It is the flowchart of the method of the present invention; Figure 2 It is the schematic diagram of the continuous scenario of multiple traffic participants at the crossroads of the present invention; Figure 3 It is the schematic diagram for calculating the projection range of obstacles and tasks of the present invention; Figure 4 It is the schematic diagram of the results of the projection of obstacles and tasks of the present invention; Figure 5 It is the schematic diagram of trajectory sampling of the present invention. Specific Embodiment

[0019] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented on the premise of the technical solution of the present invention, and the detailed implementation manners and specific operation processes are given, but the protection scope of the present invention is not limited to the following embodiments.

[0020] Unless otherwise defined, the technical terms or scientific terms involved in this application should have the ordinary meanings understood by those of ordinary skill in the technical field to which this application belongs. The "a", "one", "kind", "the" and other similar words involved in this application do not indicate a quantity limitation and may represent a single or plural number. The terms "including", "comprising", "having" and any variations thereof involved in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product or device including a series of steps or modules (units) is not limited to the listed steps or units, but may further include unlisted steps or units, or may further include other steps or units inherent to these processes, methods, products or devices. The "connection", "coupling" and other similar words involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application means two or more. The "and / or" describes the association relationship of associated objects and indicates that three relationships may exist. For example, "A and / or B" may mean: A exists alone, A and B exist simultaneously, and B exists alone. The character " / " generally represents an "or" relationship between the front and back associated objects. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific order for the objects.

[0021] In view of the technical problems existing in the prior art, the present invention provides an accelerated test method for continuous conflict scenarios of multiple traffic participants in a closed area, and its process is as Figure 1 shown.

[0022] Specifically, it includes: S1. Obtain the global path of the vehicle under test, construct a continuous conflict scenario based on the global path of the vehicle under test, and generate the global path of traffic participants using the optimized A* algorithm based on the continuous conflict scenario.

[0023] S11. Combine the high-precision map, operational design conditions (ODC), test scenario requirements, and closed-site conditions to predict the global path of the vehicle under test and represent it in the Cartesian coordinate system. The result is as Figure 2 shown. The path of the vehicle under test is to turn left at the intersection.

[0024] Among them, the ODC includes the test running state of the vehicle under test, mainly the vehicle speed information; the test scenario requirements include the topological structure of the roads to be passed by the vehicle under test, the complexity of the traffic environment, etc.

[0025] S12. Generate the global path of traffic participants.

[0026] S121. Obtain the key passing points based on the global path of the vehicle under test to form a continuous conflict scenario.

[0027] S122. Select path points using the optimized A* algorithm based on the key passing points in the continuous conflict scenario, and generate the global path of traffic participants based on the path points. The result is as Figure 2 shown. The global paths of traffic participant 1 and traffic participant 3 are the closed-loop paths before the vehicle under test turns left, and the global paths of traffic participant 2 and traffic participant 4 are the closed-loop paths after the vehicle under test turns left.

[0028] The optimized A* algorithm provided by the present invention adds the concept of path points to the A* algorithm to optimize the A* algorithm. Specifically, the concept of path points is quantified by calculating the comprehensive optimization level of the path points to be selected, and the path points are selected based on the comprehensive priority. The expression is: , , , , Among them, represents the distance of the path point n to be selected from the starting point, that is, the sum of the weights of the path; represents the estimated distance of the path point n to be selected from the end point, calculated using the Euclidean distance; represents the sum of the distances of the path point n to be selected from all key passing points, calculated using the Euclidean distance; represents the comprehensive priority of the i-th path point; represents the position of the path point n to be selected; Indicates the position of the end point; Indicates the position of the j-th critical path point; Indicates the number of critical path points.

[0029] S2. Based on the global path of the vehicle under test and the global paths of traffic participants, construct the interaction area between the vehicle under test and traffic participants. For each interaction area, use the optimization model to assign test tasks and priorities to traffic participants.

[0030] In the method provided by the present invention, for each test task, it can be jointly executed by multiple traffic participants; for each traffic participant, it can execute multiple test tasks; and in the present invention, cloud collaborative control technology is adopted to achieve task allocation for multiple traffic participants and control traffic participants according to the divided priorities.

[0031] The detailed steps are as follows: S21. Based on the global path of the vehicle under test and the global paths of traffic participants, obtain the interaction area between the vehicle under test and traffic participants, and define a corresponding test task set for the interaction area.

[0032] Among them, the interaction area is a circular area with a radius of 1 m expanded from the global path trajectory coincidence point, as shown in the circular area in Figure 2 .

[0033] [[ID=2३]]S22. Based on the optimization model of the auction algorithm and the test task set, perform test task allocation for traffic participants. This optimization allocation process is executed in the cloud, and the expression of the optimization model is: , , Among them, Indicates the net benefit when traffic participant i executes test task j, obtained based on the distance between traffic participant i and the point corresponding to test task j; N represents the number of traffic participants; Indicates the number of test tasks; Indicates the test task allocation status of traffic participant i. If It means that traffic participant i is assigned to execute test task j. If It means that the traffic participant does not execute test task j; Indicates the set of traffic participants; Indicates the upper limit value of the number of traffic participants assigned to execute test tasks.

[0034] S23. Based on the global path of the vehicle under test, obtain the order in which the vehicle under test passes through the road topology structure, and determine the priority of executing test tasks based on the order.

[0035] For each traffic participant, S3 - S5 are executed to obtain the optimal local trajectory: S3. Obtain the vehicle states of the vehicle being tested and traffic participants with priorities lower than the target traffic participant during the execution of all assigned test tasks by the target traffic participant. Based on the vehicle states, perform obstacle projection and task projection, and construct an SL graph and an ST graph. As shown in Figure 4 Figure, traffic participant 3 is selected as the target traffic participant in Figure 4 . Specifically, in Figure 4 , the driving trajectory of the vehicle being tested is to turn right into the intersection where traffic participant 4 is located. If traffic participant 3 goes straight, the interaction area between the vehicle being tested and traffic participant 3 is the circle in the figure.

[0036] Specifically, the SL graph starts from the position of the traffic participant itself, and the ST graph starts from the position of the traffic participant itself and the current time.

[0037] S31. Convert the target traffic participant from the Cartesian coordinate system to the Frenet coordinate system. In the Frenet coordinate system, decouple the trajectory of the target traffic participant into an SL graph and an ST graph in layers.

[0038] S32. Traverse the trajectories of other traffic participants to obtain the potential interaction range. Screen out traffic participants with priorities lower than the target traffic participant within the potential interaction range and project them as obstacles onto the SL graph and the ST graph respectively. Taking the scenario shown in Figure 4 as an example, assume that the priorities of traffic participant 2 and traffic participant 4 are both lower than that of traffic participant 3. Then, for traffic participant 3, traffic participant 2 and traffic participant 4 are both obstacles. After executing S321 - S322, the projection results are as shown in Figure 4 Figure.

[0039] S321. Perform geometric inflation modeling on the obstacle to obtain a geometric inflation model, and its expression is: , , , , where right front corner point of the obstacle; left front corner point of the obstacle; represents the left rear corner point of the obstacle; represents the right rear corner point of the obstacle; represents the center coordinate point of the obstacle; heading angle, that is, the included angle between the traveling direction of the obstacle and the axis; represents half of the longitudinal length of the obstacle; Represents half of the lateral length of the obstacle; Represents the longitudinal expansion coefficient; Represents the lateral expansion coefficient; S322. Convert the coordinates of the geometric expansion model to the Frenet coordinate system, and project the geometric expansion model in the Frenet coordinate system onto the SL diagram and the ST diagram.

[0040] Specifically, by converting the geometric expansion model from the Cartesian coordinate system to the Frenet coordinate system (and with its origin at the position point of the target traffic participant as the origin in this coordinate system), obtain the coordinate point values of each vertex of the obstacle, and project the obstacle onto the SL diagram depending on the coordinate point values of each vertex of the obstacle.

[0041] Obtain the relative value of the traffic participant and the target traffic participant in the longitudinal position from the geometric expansion model of the obstacle of the target traffic participant at each moment, project it onto the S axis of the ST diagram, and project the corresponding moment onto the T axis of the ST diagram to achieve the projection of the obstacle on the ST diagram.

[0042] S33. Calculate the task time domain when the target traffic participant executes each assigned test task based on the vehicle state of the vehicle under test, that is, the time when the vehicle under test reaches the interaction area corresponding to the test task.

[0043] S34. Obtain the physical sizes of the vehicle under test and the traffic participant, project the task time domain corresponding to each assigned test task onto the T axis of the ST diagram based on the physical sizes, and project the interaction area corresponding to the test task (that is, the Figure 4 area shown by the circle in the figure) onto the S axis of the SL diagram and the ST diagram to achieve task projection. Specifically, when projecting the task time domain and the interaction area, the calculation schematic diagram of the projection range is as shown in Figure 3 the figure, and the projection method of the task time domain is: , , where, represents the time when the vehicle under test reaches the interaction area, and , represents the distance from the vehicle under test to the interaction area, represents the speed of the vehicle under test; represents the projection range of the task time domain on the T axis of the ST diagram; represents half of the width of the target traffic participant (i.e., traffic participant 3).

[0044] The projection method of the interaction area is: , , wherein, represents the position where the traffic participant reaches the interaction area; represents half of the width of the vehicle under test; represents the projection range of the interaction area projected onto the S axis.

[0045] S4. Obtain the road information in the test scenario, construct the kinematic model and its constraints of the target traffic participant based on the road information, and obtain the real-time state of the target traffic participant based on the kinematic model and its constraints.

[0046] When constructing the kinematic model and constraints of the traffic participant considering the road information, the road information includes the road curvature, and its expression is: ， , , , , wherein, represents the derivative of the longitudinal distance of the traffic participant in the Frenet coordinate system; represents the speed of the traffic participant; represents the heading angle of the traffic participant; represents the curvature of the road in the Frenet coordinate system; represents the lateral distance traveled by the traffic participant in the Frenet coordinate system.

[0047] S5. Construct a multi-objective optimization function for local trajectory allocation, perform trajectory point sampling in the SL diagram and the ST diagram based on the real-time state to generate a trajectory cluster, and screen the trajectories in the trajectory cluster based on the multi-objective optimization function to obtain the optimal local trajectory.

[0048] In this step, based on the real-time state, perform trajectory point sampling of the target traffic participant in the SL diagram and the ST diagram, and combine the trajectory points to generate a trajectory cluster. The trajectory cluster includes multiple trajectories. Using the multi-objective optimization function, by calculating the multi-objective optimization function for local trajectory allocation of each trajectory, select the trajectory corresponding to the optimal value as the optimal local trajectory to achieve the optimal trajectory solution that meets the test task requirements and the safe interaction between traffic participants through local search.

[0049] Taking the scenario in Figure 4 as an example, perform scatter sampling on the SL diagram and the ST diagram based on the real-time state for the scenario in Figure 4 , and the results are as shown in Figure 5As shown by the black dots, based on the scatter sampling results, polynomial trajectories in the longitudinal and transverse directions are generated by polynomial curve fitting. The longitudinal and transverse polynomial trajectories are synthesized, and the synthesized trajectory is converted into Cartesian coordinate points, finally forming a complete trajectory. Repeating this process generates a trajectory cluster, and the effect is as shown Figure 5 by the dashed line with arrows in the figure

[0050] The detailed method for polynomial fitting of the trajectory includes: Calculating the longitudinal polynomial coefficients based on the target state and the initial state, and calculating the longitudinal trajectory based on the longitudinal polynomial coefficients: , , , , , , , , , where represents the longitudinal distance function; 、 、 、 、 and represent the longitudinal polynomial coefficients; represents time; represents the longitudinal velocity function; represents the longitudinal acceleration function; represents the initial longitudinal position; represents the initial longitudinal velocity; represents the initial longitudinal acceleration; represents the target longitudinal position at time represents the target longitudinal velocity; represents the target longitudinal acceleration.

[0051] Calculating the transverse polynomial coefficients based on the target state and the initial state, and calculating the transverse trajectory, whose expression is: , , , , , , , , , Among them, , , , , and represent the lateral polynomial coefficients; represents the initial lateral position; represents the initial lateral velocity; represents the initial lateral acceleration; represents the target lateral position at time represents the target lateral velocity; represents the target lateral acceleration; represents time.

[0052] In the present invention, the local trajectory allocation multi-objective optimization function is: , , , , , ,

[0053] , , , , Among them, represents the allocation multi-objective cost of the i-th trajectory in the trajectory cluster; represents the centripetal acceleration weight; represents the centripetal acceleration cost; represents the lateral driving cost; represents the longitudinal driving cost; represents the lateral offset weight; represents the lateral offset cost; represents the velocity at the t-th moment; represents the road curvature; Represents the total time required for the driving trajectory; Represents the lateral offset weight; Represents the lateral offset cost; Represents the lateral comfort weight; Represents the lateral comfort cost; Represents the longitudinal position at the t-th moment; Represents the lateral offset at the longitudinal position s; Represents the lateral error boundary; Represents The direction factor with the initial lateral offset; Represents the second derivative of the lateral offset at the longitudinal position s; Represents the longitudinal speed at the t-th moment; Represents the first derivative of the lateral offset at the longitudinal position s; Represents the longitudinal acceleration at the t-th moment; Represents the weight to reach the end of the trajectory; Represents the cost to reach the end of the trajectory; Represents the weight to avoid collisions; Represents the cost to avoid collisions; Represents the smoothing weight; Represents the smoothing cost; Represents the total number of trajectory points; Represents the sampling time of the j-th trajectory point; Represents the deviation of the j-th trajectory point from the desired speed; Represents the length of the i-th trajectory in the trajectory cluster; Represents the Gaussian function of the distance cost between the j-th trajectory point and the obstacle projection, and , Represents the variance of the Gaussian function, Represents the distance function between the trajectory point and the obstacle projection and , Represents that the trajectory point is within the obstacle projection determination window, Represents that the trajectory point is outside the obstacle projection determination window; Represents the longitudinal acceleration of the j-th trajectory point; Represents the maximum value of the longitudinal acceleration of the j-th trajectory point; Represents the Gaussian function of the distance cost between the j-th trajectory point and the test task, and , Represents the distance function between the trajectory point and the test task projection and , Represents the preset coefficient, Represents that the trajectory point is within the test task projection determination window, Represents that the trajectory point is in the test task projection determination window.

[0054] In addition, when generating the optimal local trajectory, the trajectory needs to be further screened. Specifically: perform collision detection and task detection on the trajectory corresponding to the optimal value. If the requirements are met, the trajectory corresponding to the optimal value is the optimal local trajectory; otherwise, delete the trajectory corresponding to the optimal value from the trajectory cluster and re-select the optimal local trajectory, so that the selected trajectory can satisfy the test task as much as possible on the premise of safety.

[0055] Among them, the collision detection is as follows: traverse all the obstacle projections in the SL graph and the ST graph, and determine whether the trajectory corresponding to the optimal value has an intersection with any obstacle projection. If there is an intersection, it is determined that there is a collision; otherwise, there is no collision.

[0056] The task detection is as follows: traverse all the test task projections in the ST graph, and determine whether the trajectory corresponding to the optimal value has an intersection with any test task projection. If there is an intersection, it is determined that the corresponding traffic participant executes the test task; otherwise, it is determined that the corresponding traffic participant does not execute the test task.

[0057] S6. In a continuous conflict scenario, the cloud sends the planned optimal local trajectory to the corresponding traffic participant. After obtaining the execution situation of all test tasks and the collision situation between the vehicle under test and the traffic participant after each traffic participant executes the optimal local trajectory, if all test tasks are executed or the vehicle under test collides with the traffic participant, the test ends, and the real driving data of the vehicle under test and the traffic participant are integrated to evaluate the test result of the vehicle under test; otherwise, return to S2, update the interaction area based on the optimal local trajectory, and re-perform the test task allocation.

[0058] In addition, the present invention also provides an electronic device, including a central processing unit (CPU), which can execute various appropriate actions and processes according to the computer program instructions stored in the read-only memory (ROM) or the computer program instructions loaded from the storage unit into the random access memory (RAM). In the RAM, various programs and data required for device operation can also be stored. The CPU, ROM, and RAM are connected to each other through a bus. The input / output (I / O) interface is also connected to the bus.

[0059] Multiple components in the device are connected to the I / O interface, including: an input unit, such as a keyboard, a mouse, etc.; an output unit, such as various types of displays, speakers, etc.; a storage unit, such as a disk, an optical disc, etc.; and a communication unit, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit allows the device to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0060] The processing unit executes the various methods and processes described above, such as methods S1 to S6. For example, in some embodiments, methods S1 to S6 may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed onto the device via the ROM and / or the communication unit. When the computer program is loaded into the RAM and executed by the CPU, one or more steps of methods S1 to S6 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute methods S1 to S6 by any other suitable means (e.g., by means of firmware).

[0061] The functions described above herein can be performed at least in part by one or more hardware logic components. By way of example and not limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGA), application specific integrated circuits (ASIC), application specific standard products (ASSP), system on a chip (SOC), complex programmable logic devices (CPLD), and the like.

[0062] The program code for implementing the methods of the present invention may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, a special purpose computer, or other programmable data processing apparatus, such that the program codes, when executed by the processor or controller, cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine, or entirely on the remote machine or server.

[0063] In the context of the present invention, a machine-readable medium may be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium may be either a machine-readable signal medium or a machine-readable storage medium. Machine-readable media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0064] As mentioned above, it is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. An accelerated test method for continuous conflict scenarios of multiple traffic participants in a closed area, characterized in that, The method described above includes: Obtain the global path of the vehicle under test, construct a continuous conflict scenario based on the global path of the vehicle under test, and generate the global path of traffic participants using the optimized A* algorithm based on the continuous conflict scenario; Construct the interaction area between the vehicle under test and traffic participants based on the global path of the vehicle under test and the global path of traffic participants, and construct an optimized model based on the auction algorithm. For each of the interaction areas, use the optimized model to assign test tasks and priorities to traffic participants; For each traffic participant, perform the following steps to obtain the optimal local trajectory: Obtain the vehicle states of the vehicle under test and traffic participants with priorities lower than the target traffic participant during the execution of all assigned test tasks by the target traffic participant, project obstacles based on the vehicle states of traffic participants with priorities lower than the target traffic participant, project tasks based on the vehicle state of the vehicle under test, and construct the SL graph and ST graph of the target traffic participant; Obtain the road information in the test scenario, construct the kinematic model and its constraints of the target traffic participant based on the road information, and obtain the real-time state of the target traffic participant based on the kinematic model and its constraints; the road information includes road curvature; Construct a multi-objective optimization function for local trajectory allocation, sample trajectory points of the target traffic participant in the SL graph and ST graph based on the real-time state, and combine the trajectory points to generate a trajectory cluster. The trajectory cluster includes multiple trajectories. Calculate the multi-objective optimization function for local trajectory allocation of each trajectory, and select the trajectory corresponding to the optimal value as the optimal local trajectory; Obtain the execution status of all test tasks and the collision situation between the vehicle under test and traffic participants after each traffic participant executes the optimal local trajectory. If all test tasks are executed or the vehicle under test collides with traffic participants, end the test; otherwise, update the interaction area based on the optimal local trajectory and re-allocate test tasks.

2. The accelerated test method for continuous conflict scenarios of multiple traffic participants in a closed area according to claim 1, wherein The method for generating the global path of traffic participants is: Obtain key waypoints based on the global path of the vehicle under test, select path points using the optimized A* algorithm based on the key waypoints, and generate the global path of traffic participants based on the path points; Among them, the method for selecting path points is to calculate the comprehensive priority of candidate path points and select path points based on the comprehensive priority. Its expression is: ， ， ， ， Among them, represents the distance of the to-be-selected path point n from the starting point; represents the estimated distance of the to-be-selected path point n from the end point; represents the sum of the distances of the to-be-selected path point n from all the key path points; represents the comprehensive priority of the i-th path point; represents the position of the to-be-selected path point n; represents the position of the end point; represents the position of the j-th key path point; represents the number of key path points.

3. The accelerated test method for a continuous conflict scenario of multiple traffic participants in a closed area according to claim 1, characterized in that, The method for allocating test tasks to traffic participants and their priorities when executing the test tasks is: Obtain the interaction area between the vehicle under test and traffic participants based on the global path of the vehicle under test and the global path of traffic participants, and define a set of test tasks for the interaction area; Based on the optimized model of the auction algorithm and the set of test tasks, perform traffic participant test task allocation. The expression of the optimized model is: ， ， Among them, represents the net benefit when traffic participant i performs test task j, obtained based on the distance of traffic participant i from the corresponding point of test task j; N represents the number of traffic participants; represents the number of test tasks; represents the test task allocation status of traffic participant i. If then it means that traffic participant i is assigned to perform test task j. If then it means that the traffic participant does not perform test task j; represents the set of traffic participants; represents the upper limit value of the number of traffic participants assigned to perform test tasks; Based on the global path of the vehicle under test, obtain the order of the vehicle under test passing through the road topology structure, and determine the priority of executing test tasks based on the order.

4. The accelerated test method for continuous conflict scenarios of multiple traffic participants in a closed area according to claim 1, characterized in that, The method for constructing the SL graph and ST graph of the target traffic participant includes: Convert the target traffic participant from the Cartesian coordinate system to the Frenet coordinate system. In the Frenet coordinate system, layer and decouple the trajectory of the target traffic participant into an SL graph and an ST graph; Traverse the trajectories of other traffic participants to obtain the potential interaction range. Screen the traffic participants with a lower priority than the target traffic participant within the potential interaction range and project them onto the SL graph and the ST graph as obstacles respectively; Calculate the task time domain when the target traffic participant executes each assigned test task based on the vehicle state of the vehicle under test, that is, the time when the vehicle under test reaches the interaction area corresponding to the test task; Obtain the physical sizes of the vehicle under test and the traffic participants. Project the task time domain corresponding to each assigned test task onto the T axis of the ST graph based on the physical sizes, and project the interaction area corresponding to the test task onto the S axis of the SL graph and the ST graph to achieve task projection.

5. The accelerated test method for a continuous conflict scenario of multiple traffic participants in a closed area according to claim 4, wherein, The projection method of the obstacle is as follows: Perform geometric dilation modeling on the obstacle to obtain a geometric dilation model, and its expression is: ， ， ， ， Among them, The right front corner point of the obstacle; The left front corner point of the obstacle; Indicates the left rear corner point of the obstacle; Indicates the right rear corner point of the obstacle; Indicates the center coordinate point of the obstacle; The heading angle, that is, the included angle between the traveling direction of the obstacle and the axis; Indicates half of the longitudinal length of the obstacle; Indicates half of the transverse length of the obstacle; Indicates the longitudinal expansion coefficient; Indicates the transverse expansion coefficient; Convert the coordinates of the geometric dilation model to the Frenet coordinate system, and project the geometric dilation model in the Frenet coordinate system onto the SL graph and the ST graph.

6. The accelerated test method for continuous conflict scenarios of multiple traffic participants in a closed area according to claim 4, wherein The projection method of the task time domain is as follows: ， ， Among them, represents the time when the vehicle under test arrives at the interaction area, and , represents the distance from the vehicle under test to the interaction area, represents the speed of the vehicle under test; represents the projection range of the task time domain on the T-axis of the ST diagram; represents half of the width of the target traffic participant.

7. The accelerated test method for continuous conflict scenarios of multiple traffic participants in a closed area according to claim 4, characterized in that The projection method of the interaction area is as follows: ， ， Among them, represents the position where the traffic participant arrives at the interaction area; represents half of the width of the vehicle under test; represents the projection range of the interaction area projected onto the S axis.

8. The accelerated test method for continuous conflict scenarios of multiple traffic participants in a closed area according to claim 1, wherein The kinematic model and its constraints are: ， ， ， ， ， Among them, represents the derivative of the longitudinal distance of a traffic participant in the Frenet coordinate system; represents the speed of a traffic participant; represents the heading angle of a traffic participant; represents the curvature of the road in the Frenet coordinate system; represents the lateral distance traveled by a traffic participant in the Frenet coordinate system.

9. The accelerated test method for a continuous conflict scenario of multiple traffic participants in a closed area according to claim 1, characterized in that, The multi-objective optimization function for local trajectory allocation is: ， ， ， ， ， ， ， ， ， ， ， Among them, represents the allocated multi-objective cost of the i-th trajectory in the described trajectory cluster; represents the centripetal acceleration weight; represents the centripetal acceleration cost; represents the lateral driving cost; represents the longitudinal driving cost; represents the lateral offset weight; represents the lateral offset cost; represents the speed at the t-th moment; represents the road curvature; represents the total time required for the driving trajectory; represents the lateral offset weight; represents the lateral offset cost; represents the lateral comfort weight; represents the lateral comfort cost; represents the longitudinal position at the t-th moment; represents the lateral offset at the longitudinal position s; represents the lateral error boundary; represents the direction factor with the initial lateral offset; represents the second derivative of the lateral offset at the longitudinal position s; represents the longitudinal speed at the t-th moment; represents the first derivative of the lateral offset at the longitudinal position s; represents the longitudinal acceleration at the t-th moment; represents the weight for reaching the end point of the trajectory; represents the cost for reaching the end point of the trajectory; represents the weight for collision avoidance; represents the cost for collision avoidance; represents the smoothing weight; represents the smoothing cost; represents the total number of trajectory points; represents the sampling time of the trajectory point j; represents the deviation of the trajectory point j from the desired speed; represents the length of the i-th trajectory in the trajectory cluster; represents the Gaussian function of the distance cost between the trajectory point j and the obstacle projection; represents the longitudinal acceleration of the trajectory point j; represents the maximum value of the longitudinal acceleration of the trajectory point j; represents the Gaussian function of the distance cost between the trajectory point j and the test task.

10. The accelerated test method for continuous conflict scenarios of multiple traffic participants in a closed area according to claim 1, wherein The method for obtaining the optimal local trajectory further includes: performing collision detection and task detection on the trajectory corresponding to the optimal value. If the requirements are met, the trajectory corresponding to the optimal value is the optimal local trajectory; otherwise, delete the trajectory corresponding to the optimal value from the trajectory cluster and re-select the optimal local trajectory; Among them, the collision detection is: traverse all obstacle projections in the SL graph and the ST graph, and determine whether the trajectory corresponding to the optimal value intersects with any of the obstacle projections. If so, it is determined that there is a collision, otherwise there is no collision; The task detection is: traverse all test task projections in the ST graph, and determine whether the trajectory corresponding to the optimal value intersects with any of the test task projections. If so, it is determined that the corresponding traffic participant executes the test task, otherwise it is determined that the corresponding traffic participant does not execute the test task.

Images