Simulation test method and device

Abstract

The invention discloses a simulation test method and device, and specifically discloses a method for determining an abnormal driving event occurring in the driving process of specified equipment from acquired driving data of the specified equipment in the driving process, determining at least one to-be-executed simulation test task corresponding to the abnormal driving event; for each simulation test task, determining test data corresponding to the simulation test task from the driving data, determining scene feature data corresponding to the simulation test task according to the test data, and after determining that the scene feature data matched with the scene feature data does not exist in a preset simulation database, storing the test data corresponding to the simulation test task and the scene feature data in the simulation database, and constructing a corresponding simulation scene according to the test data corresponding to the simulation test task to execute the simulation test task. In this way, the artificial design cost can be reduced, and meanwhile the simulated test scene is more real.

Description

technical field [0001] This specification relates to the field of Internet technology, and in particular to a method and device for simulation testing. Background technique [0002] In the field of unmanned driving technology, it is an essential part of the algorithm iteration process to simulate and test the decision-making algorithm by generating simulation test scenarios through simulation software. At present, when performing simulation tests on decision-making algorithms, the number of test scenarios stored in the simulation database is limited. In order to enable the simulation testing of decision-making algorithms to effectively cover various test scenarios, the research and development institutions of decision-making algorithms must purchase additional corresponding test scenario library, which will inevitably lead to an increase in economic costs. However, if the test scenario library is not purchased, and the testers design and build the corresponding test scenari...

AI Technical Summary

Problems solved by technology

At present, when performing simulation tests on decision-making algorithms, the number of test scenarios stored in the simulation database is limited. In order to enable the simulation testing of decision-making algorithms to effectively cover various test scenarios, the research and development institutions of decision-making algorithms must purchase additional corresponding test scenario library, which will inevitably lead to an increase in economic costs

However, if the test scenario library is not purchased and the testers design and build the corresponding test scenarios by themselves, it will take a lot of labor costs

Furthermore, the test scenarios designed and built by testers based on their own experience are often limited by their own thinking. Compared with the real driving environment of unmanned driving equipment, the test scenarios built in this way are relatively ideal and cannot cover well Problems of unmanned driving equipment in the real driving process

Images