Automatic driving simulation multi-platform linkage test method based on offline task allocation

Abstract

The invention belongs to the technical field of automatic driving testing, and particularly relates to an automatic driving simulation multi-platform linkage testing method based on offline task allocation. Comprising the steps that firstly, multi-platform linkage test offline task allocation is subjected to mathematical expression, and key element definitions including test duration element definitions and test load element definitions are clarified; 2, establishing test task allocation constraints including a test precision constraint, a test priority constraint and a test load constraint; and 3, solving a task allocation result by using a particle swarm algorithm so as to obtain specific tasks of different test platforms. According to the method, linkage testing of the multiple simulation testing platforms is achieved through testing task distribution, and by setting testing process distribution constraint and solving, the linkage testing method can be achieved when the multiple testing platforms exist in the automatic driving system; the test platform precision is considered in the task allocation process, and the multi-platform test capability can be utilized, so that the test efficiency can be improved while the test precision is ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of autonomous driving testing, and specifically is a multi-platform linkage testing method for autonomous driving simulation based on offline task allocation. Background Art

[0002] As the technology of autonomous vehicles continues to mature, its mass production requires a scientific and complete test and evaluation system as support. At present, scenario-based simulation testing has become an important support for solving the problem of autonomous vehicle performance verification. However, autonomous vehicle simulation testing involves different types of test platforms. Considering the problem of dimensional explosion caused by the large number of scene elements, how to coordinate the advantages of simulation test platforms to establish a multi-simulation platform linkage test method is an important means to improve test efficiency and ensure test accuracy. Although the research on multi-platform linkage testing has begun, the test task allocation is mostly concentrated on the simple allocation of scene elements or directly integrating multiple platforms. The former cannot fully utilize the accuracy advantages of each test platform, while the latter needs to test all test items, which is prone to dimensional explosion. To this end, analyzing the technical advantages of each simulation test platform and establishing a multi-platform linkage test method has become an important part of autonomous vehicle simulation testing. Summary of the invention

[0003] In order to solve the above technical problems, the present invention provides an autonomous driving simulation multi-platform linkage test method based on offline task allocation, which realizes linkage testing of multiple simulation test platforms through test task allocation. The method can solve the linkage test method when there are multiple test platforms in the autonomous driving system by setting test process allocation constraints and solving them; since the test platform accuracy is taken into consideration in the task allocation process and the multi-platform testing capability can be utilized, the test efficiency can be improved while ensuring the test accuracy.

[0004] The technical solution of the present invention is described as follows in conjunction with the accompanying drawings: A multi-platform linkage test method for autonomous driving simulation based on offline task allocation includes the following steps: Step 1: mathematically express the offline task allocation of multi-platform linkage test and clarify the definition of key elements, including the definition of test duration element and test load element; Step 2: Establish test task allocation constraints, including test accuracy constraints, test priority constraints, and test load constraints; Step 3: Use the particle swarm algorithm to solve the task allocation results to obtain specific tasks for different test platforms.

[0005] Furthermore, the specific method of step one is as follows: 11) Describe the multi-platform test task allocation for autonomous vehicles; Self-driving cars have m Specific scenario parameter combinations to be tested U , represented by { U 1, U 2, ..., U m For multiple simulation test platforms, the test task allocation process will be m The test tasks are assigned to the simulation test platform that needs linkage test. P , represented by { P s , P c , P m , ……}; For each test task, it contains a specific scenario parameter combination U i , a distributed simulation testbench P j , the corresponding test time t ij ; The total number of test tasks is m or less than m ,On this basis, the time for different simulation test platforms to execute ,test tasks is considered, and the test load estimation value is ,established for the test tasks; 12) Define the test task allocation matrix; Defined test task allocation matrix T As shown in formula (1): (1) In the formula, t s , t c , t m The subscripts are 1, 2, 3, ..., respectively, and are the time consumed by different types of simulation test platforms to execute a test task. q Number the test tasks executed by different test platforms; q The number of test tasks assigned to the test platform with the largest number of test tasks; i In the matrix j When there is no actual test task in a test process, the corresponding element in the matrix t i,j =0; For a logical scenario described by parameter space, a large number of test processes for specific tests are obtained. The final test time is shown in formula (2): (2) The final objective function of the entire testing process is thus established: (3) 13) Define the test duration; The test duration constraint refers to the time required for different test platforms to test a single scenario, that is, t s , t c , t m , …… ; For different simulation test platforms, when performing the same test task, there is a certain deviation in the test time. When the test task running time is set to t set During the automated test, the actual test duration of the digital simulation test platform and the hardware-in-the-loop test platform will be longer than the set scenario duration due to scenario calling and parameter reset. The duration delays caused by the platform automation performance are t s_y , t c_y , t m_y , …… , the time consumption of different simulation platform test tasks is shown in formula (4): (4) In the formula, For the i The expected time it takes for a simulation test platform to execute a test task; The duration of the test task definition when the i-th simulation test platform executes the test task; The delay of the testing process is caused by the performance limitation of the platform; 14) Define the test load; The test load is the probability that the task will be actually tested by the platform; the test load λ There are two types of parameters: continuous parameters and enumeration parameters. The digital representation is as follows: (5) In the formula, Test loads defined for continuous class parameters; Test loads defined for discrete class parameters.

[0006] Furthermore, the specific method of step 14) is as follows: a. Define continuous class parameter load; Definition i Test load factor for consecutive parameter combinations λ c_i for: (6) (7) In the formula, δ It is the adjustment coefficient that keeps shrinking as the test is iterated; n c is the number of dimensions of the continuous scene parameters; d i_min For the i The distance between continuous parameter combinations is the shortest normalized distance between the test results and the opposite continuous parameter combinations; q It is the maximum value of the test task on a single platform; k t The number of experiments that have been conducted on the platform; d i_min The calculation method uses the maximum and minimum value normalization method, as shown below: (8) In the formula, p j The continuous parameter combination calculated in j The value of the dimension; p min_j The point opposite to the experimental result closest to the calculated parameter combination is j The value of the dimension; and The logical scene parameter space is j The maximum and minimum values ​​of the dimension; b. Define enumeration class parameter test load; Definition i Class Enumeration Class Parameter Test Load λ d_i The calculation is as follows: (9) In the formula, w i For the i Class enumeration class parameter difficulty coefficient; w near The difficulty coefficient is i Class enumeration class parameter test difficulty closest to the tested parameter combination difficulty coefficient; w worst is the maximum difficulty coefficient of the enumeration parameter combination; if w i and w near Equal and w near The test result is safe, define the λ d_i is 0.5.

[0007] Furthermore, the specific method of step 2 is as follows: 21) Define test accuracy constraints; The accuracy threshold constraint means that the test task can only be assigned to the test platform that meets the test accuracy requirements. If there is a test task for which all test platforms do not meet the test accuracy requirements, the test task will not participate in the platform task allocation process, as shown in formula (10): (10) In the formula, For the The test platform is aimed at The accuracy results of a specific test scenario; For the A specific test scenario is assigned to the The probability of a test platform; 22) Define test priority constraints; The test priority constraint means that when the test platform that meets the test task accuracy requirements includes different test platforms at the same time, the test task is preferentially allocated to the test platform with higher test accuracy. This part of the constraint is determined by defining the time reduction coefficient. After adding the time reduction coefficient, the multi-platform test task allocation process matrix is ​​shown in formula (11): (11) in, (12) In the formula, For the The test platform's duration reduction factor for the specific scenario parameters; 23) Define test load balancing constraints; The test load balancing constraints are expressed as: (13) In the formula, To test the load; The test load is The test tasks are assigned to the test platform the number of For all test loads The average value of the tasks assigned to all platforms, subscript Represents different simulation test platforms.

[0008] Furthermore, the specific method of step three is as follows: 31) The ultimate goal of test task allocation is to find the optimal solution of the final objective function while satisfying multiple constraints, and to equate the final goal to a multi-objective optimization process; 32) Use the improved particle swarm algorithm to accelerate the solution of the multi-objective optimization process; when using the particle swarm algorithm to solve the task allocation, the search for the optimal solution is completed by the movement of particles in the solution space; the update of speed and position in the particle swarm algorithm is shown in formulas (14)-(15), and the optimal solution is finally obtained through continuous iteration of speed and position; (14) (15) In the formula, in the formula, For the Round Iteration Particle exist The location of the dimension; For the Round Iteration Particle exist The location of the dimension; For the Round Iteration Particle exist The speed of the dimension; For the Round Iteration Particle exist The speed of the dimension; is the inertia weight; and is the learning factor; and It is a random number, ranging from 0 to 1; For the Wheel Particles exist The optimal position of the dimension; For the All particles in the wheel The optimal position of the dimension.

[0009] The beneficial effects of the present invention are: The present invention realizes linkage testing of multiple simulation test platforms through test task allocation; the method can solve the linkage testing method when there are multiple test platforms in the automatic driving system by setting test process allocation constraints and solving them. Since the test platform accuracy is taken into account in the task allocation process and the multi-platform testing capability can be used, the test efficiency can be improved while ensuring the test accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 It is a schematic diagram of the process of the present invention; Figure 2 Schematic diagram of convergence results when allocating test tasks for 300 test tasks; Figure 3 Schematic diagram of convergence results when allocating test tasks for 500 test tasks; Figure 4 Schematic diagram of convergence results when allocating test tasks for 800 test tasks; Figure 5 Schematic diagram of the convergence results when allocating test tasks for 1000 test tasks. DETAILED DESCRIPTION

[0012] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0013] Example 1: See Figure 1 This embodiment provides a multi-simulation platform linkage test method for autonomous driving vehicle simulation testing, including the following steps: Step 1: Mathematically express the offline task allocation of multi-platform linkage test and clarify the definition of key elements, including the definition of test duration element and test load element, as follows: 11) Describe the multi-platform test task allocation for autonomous vehicles; For the multi-platform test task allocation of autonomous driving vehicles, it can be described as the following process: the autonomous driving vehicle has m Specific scenario parameter combinations to be tested U , represented by { U 1, U 2, ..., U m For multiple simulation test platforms, the test task allocation process will be m The test tasks are assigned to the simulation test platform that needs linkage test. P , represented by { P s ,P c , P m , ……}; For each test task, it contains a specific scenario parameter combination U i , a distributed simulation testbench P j , the corresponding test time t ij ; The total number of test tasks is m or less than m ,On this basis, the time for different simulation test platforms to execute ,test tasks is considered, and the test load estimation value is ,established for the test tasks; 12) Define the test task allocation matrix; Defined test task allocation matrix T As shown in formula (1): (1) In the formula, t s , t c , t m The subscripts are 1, 2, 3, ..., respectively, and are the time consumed by different types of simulation test platforms to execute a test task. q Number the test tasks executed by different test platforms; q The number of test tasks assigned to the test platform with the largest number of test tasks; i In the matrix j When there is no actual test task in a test process, the corresponding element in the matrix t i,j =0 (some platforms behind matrix 1 may have no test tasks, and their test time is 0); For a logical scenario described by parameter space, a large number of test processes for specific tests are obtained. The final test time is shown in formula (2): (2) The final objective function of the entire testing process is thus established: (3) 13) Define the test duration; The test duration constraint refers to the time required for different test platforms to test a single scenario, that is, t s , t c , t m ,…… ; For different simulation test platforms, when performing the same test task, there is a certain deviation in the test time. When the test task running time (the duration of a single specific scenario from the set start to the end) is set to t set During the automated test, the actual test duration of the digital simulation test platform and the hardware-in-the-loop test platform will be longer than the set scenario duration due to scenario calling and parameter resetting. The duration delays caused by the platform automation performance are t s_y , t c_y , t m_y , …… , the time consumption of different simulation platform test tasks is shown in formula (4): (4) In the formula, For the i The expected time it takes for a simulation test platform to execute a test task; The duration of the test task definition when the i-th simulation test platform executes the test task; The delay of the testing process is caused by the performance limitation of the platform; 14) Define the test load; For different test tasks, there are certain differences in their test loads. The test load can be understood as the probability that the task will be actually tested by the platform. Since the current simulation test platform mostly uses accelerated testing methods, not all test tasks will be tested. These tasks that will not be tested can be understood as having a low test load. In order to avoid task allocation errors caused by experience, the test loads of different specific test scenarios are not considered when the test tasks are initially allocated. This part needs to be considered when the test tasks are redistributed in the future with certain test results. It should be noted that when a single test platform conducts specific tests, the test scenario elements with high test loads are not necessarily tested first. This part is only to ensure that each platform takes approximately the same amount of time at the end of the final test. Test load λ There are two types of considerations: continuous parameters and enumeration parameters. For calculating a set of specific scenario parameters to be tested, first determine the test value of the continuous scenario parameter part of the parameter. If the continuous scenario parameter part of the set of parameters has not been tested, then directly use the test load of this part of the continuous scenario parameters to represent the test load of the entire specific scenario parameter combination. If the continuous scenario parameter part of the set of parameters has been tested, then consider using enumeration parameters to calculate the test load of the set of specific scenario parameters. The above analysis process is converted into a digital representation as follows: (5) In the formula, Test loads defined for continuous class parameters; Test loads defined for discrete class parameters.

[0014] Specifically, a. define continuous class parameter load; Definition i Test load factor for consecutive parameter combinations λ c_i for: (6) (7) In the formula, δ It is the adjustment coefficient that keeps shrinking as the test is iterated; n c is the number of dimensions of the continuous scene parameters; d i_min For the i The distance between continuous parameter combinations is the shortest normalized distance between the test results and the opposite continuous parameter combinations; q It is the maximum value of the test task on a single platform; k t is the number of tests that have been conducted on the current platform; the reason why the constant value in formula (7) is 3 is that the existing accelerated test algorithm generally consumes about 30% of the computing power to determine the performance boundary of the system under test. Therefore, one third of the total number of tests is used as the basis for parameter adjustment in this formula; d i_min The calculation method uses the maximum and minimum value normalization method, as shown below: (8) In the formula, p j The continuous parameter combination calculated in j The value of the dimension; p min_j The point opposite to the experimental result closest to the calculated parameter combination is j The value of the dimension; and The logical scene parameter space is j The maximum and minimum values ​​of the dimension; it should be noted that p min_j It is not the distance between the measured point and j The point with the shortest combined distance to the test result is not the point with the closest dimension to the test result.

[0015] b. Define enumeration parameter test load; For the enumeration parameter test load, it mainly considers the difficulty of the untested parameter combination, which mainly includes two aspects: if the currently tested continuous parameter combination is a dangerous situation, the higher the difficulty of the untested enumeration parameters, the lower the test load. This is because if the tested algorithm has already failed in a simple scenario, it is likely to have an accident in a complex scenario; if the currently tested continuous parameter combination is a safe situation, the higher the difficulty of the untested enumeration parameters, the higher the test load. This is because although the tested algorithm has not had an accident in a simple scenario, it may have an accident in a complex enumeration parameter combination. According to the above analysis process, the first i Class Enumeration Class Parameter Test Load λ d_i The calculation is as follows: (9) In the formula, w i For the i Class enumeration class parameter difficulty coefficient; w near The difficulty coefficient is i Class enumeration class parameter test difficulty closest to the tested parameter combination difficulty coefficient; w worst is the maximum difficulty coefficient of the enumeration parameter combination; if w i and w near Equal and w near The test result is safe, define the λ d_i is 0.5.

[0016] Step 2: Establish test task allocation constraints, including test accuracy constraints, test priority constraints, and test load constraints, as follows: 21) Define test accuracy constraints; The accuracy threshold constraint means that the test task can only be assigned to the test platform that meets the test accuracy requirements. If there is a test task for which all test platforms do not meet the test accuracy requirements, the test task will not participate in the platform task allocation process, as shown in formula (10): (10) In the formula, F r_i_j For the i The test platform is aimed at j The accuracy results of a specific test scenario; P i_j For the j A specific test scenario is assigned to the iThe probability of a test platform; 22) Define test priority constraints; The test priority constraint means that when the test platform that meets the test task accuracy requirements includes different test platforms at the same time, the test task is preferentially allocated to the test platform with higher test accuracy. This part of the constraint is determined by defining the time reduction coefficient. After adding the time reduction coefficient, the multi-platform test task allocation process matrix is ​​shown in formula (11): (11) in, (12) In the formula, For the The test platform's duration reduction factor for the specific scenario parameters; 23) Define test load balancing constraints; For test task allocation, in order to avoid allocating all test tasks with high test load to a single test platform under the goal of obtaining the optimal test time, it is necessary to add test load balancing constraints to ensure the balance of different test platforms during the test load process. The test load balancing constraints can be expressed as: (13) In the formula, To test the load; The test load is The test tasks are assigned to the test platform the number of For all test loads The average value of the tasks assigned to all platforms, subscript Represents different simulation test platforms.

[0017] Step 3: Use the particle swarm algorithm to solve the task allocation results to obtain specific tasks for different test platforms. The specific method is as follows: 31) The ultimate goal of test task allocation is to find the optimal solution of formula (3) while satisfying multiple constraints. The above process can be equivalent to a multi-objective optimization process; 32) Use the improved particle swarm algorithm to accelerate the process. When using the particle swarm algorithm to solve the task allocation, each particle is equivalent to completing a test task allocation, and each dimension in the particle can be considered as the allocation result of the corresponding task. By adjusting the allocation strategy differently, the test time can be optimized. The particle swarm algorithm is an optimization algorithm based on swarm intelligence. The algorithm regards each solution in the solution space as a "particle" and completes the search for the optimal solution through the movement of particles in the solution space. In this process, the most important changes include the position of the particles. x ijand speed v ij Among them, the position represents the current solution of the particle, which represents the allocation result of a set of test tasks, and the speed represents the movement rate of the particle in the solution space, which represents the adjustment status of the test tasks. The update of speed and position in the traditional particle swarm algorithm is shown in formulas (14)-(15). Through the continuous iteration of speed and position, the optimal solution is finally obtained. (14) (15) In Chinese, For the Round Iteration Particle exist The location of the dimension; For the Round Iteration Particle exist The location of the dimension; For the Round Iteration Particle exist The speed of the dimension; For the Round Iteration Particle exist The speed of the dimension; is the inertia weight; and is the learning factor; and It is a random number, ranging from 0 to 1; For the Wheel Particles exist The optimal position of the dimension; For the All particles in the wheel The optimal position of the dimension.

[0018] Example 2: This example randomly generates 4 groups of 300, 500, 800, and 1000 specific test tasks and corresponding platform test accuracy results, uses the proposed test task allocation method to allocate test tasks on multiple simulation platforms, and calculates the corresponding overall test process time. All test task loads and platform types that meet the accuracy requirements are randomly generated, and the basic duration of each test task is set to 30s. Some of the generated test task requirements and the corresponding platform simulation results are shown in Table 1.

[0019] Table 1 Test tasks and platform realism data

[0020] The optimal test duration of each iteration of the particle swarm algorithm in the four-group test task allocation process is as follows: Figure 2-Figure 5 As shown. Figure 2-Figure 5 It can be seen from the results in that the algorithm shows good convergence results when allocating different numbers of test tasks, and can effectively reduce the overall task time when different test platforms are collaboratively tested.

[0021] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-platform linkage test method for autonomous driving simulation based on offline task allocation, characterized in that: The following steps are involved: Step 1: mathematically express the offline task allocation of multi-platform linkage test and clarify the definition of key elements, including the definition of test duration element and test load element; Step 2: Establish test task allocation constraints, including test accuracy constraints, test priority constraints, and test load constraints; Step 3: Use the particle swarm algorithm to solve the task allocation results to obtain specific tasks for different test platforms.

2. The method for multi-platform linkage testing of autonomous driving simulation based on offline task allocation according to claim 1, characterized in that: The specific method of step one is as follows: 11) Describe the multi-platform test task allocation for autonomous vehicles; Self-driving cars have m Specific scenario parameter combination to be tested U , represented by { U 1, U 2, ..., U m For multiple simulation test platforms, the test task allocation process will be m The test tasks are assigned to the simulation test platform that needs linkage test. P , represented by { P s , P c , P m , …… }; For each test task, it contains a specific scenario parameter combination U i , a distributed simulation testbench P j , the corresponding test time t ij ; The total number of test tasks is m or less than m ,On this basis, the time for different simulation test platforms to execute ,test tasks is considered, and the test load estimation value is ,established for the test tasks; 12) Define the test task allocation matrix; Defined test task allocation matrix T As shown in formula (1): （1） In the formula, t s , t c , t m The subscripts are 1, 2, 3, ..., respectively, and are the time consumed by different types of simulation test platforms to execute a test task. q Number the test tasks executed by different test platforms; q The number of test tasks assigned to the test platform with the largest number of tasks; When the test platform i In the matrix j When there is no actual test task in a test process, the corresponding element in the matrix t i,j =0; For a logical scenario described by parameter space, a large number of test processes for specific tests are obtained. The final test time is shown in formula (2): （2） The final objective function of the entire testing process is thus established: （3） 13) Define the test duration; The test duration constraint refers to the time required for different test platforms to test a single scenario, that is, t s , t c , t m , …… ; for For different simulation test platforms, when performing the same test task, the test time has certain deviations. When the test task running time is set to t set During the automated test, the actual test duration of the digital simulation test platform and the hardware-in-the-loop test platform will be longer than the set scenario duration due to scenario calling and parameter reset. The duration delays caused by the platform automation performance are t s_y , t c_y , t m_y , …… , the time consumption of different simulation platform test tasks is shown in formula (4): （4） In the formula, For the i The expected time it takes for a simulation test platform to execute a test task; The duration of the test task definition when the i-th simulation test platform executes the test task; The delay of the testing process is caused by the performance limitation of the platform; 14) Define the test load; The test load is the probability that the task will be actually tested by the platform; the test load λ There are two types of parameters: continuous parameters and enumeration parameters. The digital representation is as follows: （5） In the formula, Test loads defined for continuous class parameters; Test loads defined for discrete class parameters.

3. The method for multi-platform linkage testing of autonomous driving simulation based on offline task allocation according to claim 2 is characterized in that: The specific method of step 14) is as follows: a. Define continuous class parameter load; Definition i Test load factor for consecutive parameter combinations λ c_i for: （6） （7） In the formula, δ It is the adjustment coefficient that keeps shrinking as the test is iterated; n c is the number of dimensions of the continuous scene parameters; d i_min For the i The distance between continuous parameter combinations is the shortest normalized distance between the test results and the opposite continuous parameter combinations; q It is the maximum value of the test task on a single platform; k t The number of experiments that have been conducted on the platform. d i_min The calculation method uses the maximum and minimum value normalization method, as shown below: （8） In the formula, p j The continuous parameter combination calculated in j The value of the dimension; p min_j The point opposite to the experimental result closest to the calculated parameter combination is j The value of the dimension; and The logical scene parameter space is j The maximum and minimum values ​​of the dimension; b. Define enumeration parameter test load; Definition i Class Enumeration Class Parameter Test Load λ d_i The calculation is as follows: （9） In the formula, w i For the i Class enumeration class parameter difficulty coefficient; w near The difficulty coefficient is i Class enumeration class parameter test difficulty closest to the tested parameter combination difficulty coefficient; w worst is the maximum difficulty coefficient of the enumeration parameter combination; if w i and w near Equal and w near The test result is safe, define the λ d_i is 0.

5.

4. The method for multi-platform linkage testing of autonomous driving simulation based on offline task allocation according to claim 1, characterized in that: The specific method of step 2 is as follows: 21) Define test accuracy constraints; The accuracy threshold constraint means that the test task can only be assigned to the test platform that meets the test accuracy requirements. If there is a test task for which all test platforms do not meet the test accuracy requirements, the test task will not participate in the platform task allocation process, as shown in formula (10): （10） In the formula, For the The test platform is aimed at The accuracy results of a specific test scenario; For the A specific test scenario is assigned to the The probability of a test platform; 22) Define test priority constraints; The test priority constraint means that when the test platform that meets the test task accuracy requirements includes different test platforms at the same time, the test task is preferentially allocated to the test platform with higher test accuracy. This part of the constraint is determined by defining the time reduction coefficient. After adding the time reduction coefficient, the multi-platform test task allocation process matrix is ​​shown in formula (11): （11） in, （12） In the formula, For the The test platform's duration reduction factor for the specific scenario parameters; 23) Define test load balancing constraints; The test load balancing constraints are expressed as: （13） In the formula, To test the load; The test load is The test tasks are assigned to the test platform the number of For all test loads The average value of the tasks assigned to all platforms, subscript Represents different simulation test platforms.

5. The method for multi-platform linkage testing of autonomous driving simulation based on offline task allocation according to claim 1, characterized in that: The specific method of step three is as follows: 31) The ultimate goal of test task allocation is to find the optimal solution of the final objective function while satisfying multiple constraints, and to equate the final goal to a multi-objective optimization process; 32) Use the improved particle swarm algorithm to accelerate the solution of the multi-objective optimization process; when using the particle swarm algorithm to solve the task allocation, the search for the optimal solution is completed by the movement of particles in the solution space; the update of speed and position in the particle swarm algorithm is shown in formulas (14)-(15), and the optimal solution is finally obtained through continuous iteration of speed and position; （14） （15） In the formula, For the Round Iteration Particle exist The location of the dimension; For the Round Iteration Particle exist The location of the dimension; For the Round Iteration Particle exist The speed of the dimension; For the Round Iteration Particle exist The speed of the dimension; is the inertia weight; and is the learning factor; and It is a random number, ranging from 0 to 1; For the Wheel Particles exist The optimal position of the dimension; For the All particles in the wheel The optimal position of the dimension.

Images