Self-driving car testing system and method

Abstract

The invention discloses a test system and a method for a self-driving car. The system comprises a test site, a simulation analysis platform, and a self-driving car to be tested. The simulation analysis platform is used for constructing an environment virtual model and a vehicle virtual model, which are visualized, and comparing the real information and the virtual information. The simulation analysis platform is further used for outputting a control instruction to the hardware simulation bench according to the virtual information, to perform the power simulation, while under the condition thatthe real information matches the virtual information. The simulation analysis platform compares the simulation execution result with the execution decision of a real car, and outputs the test analysis data of the self-driving car to be detected accordingly. According to the test system and the method for the self-driving car, which achieves the comprehensive verification of the self-driving aboutthe ability of sense, make decisions, execute, and the like, by combining the virtual simulation and the real site test.

Description

technical field [0001] The invention relates to the field of automatic driving, in particular to an automatic driving vehicle testing system and method. Background technique [0002] The automatic driving system perceives the surrounding environment of the vehicle based on the environment perception technology, and according to the information obtained by the perception, controls the steering and speed of the vehicle autonomously through the decision-making controller, so that the vehicle can drive safely and reliably, and reach the predetermined destination. [0003] The test for self-driving cars is mainly based on simulation tests, but the perception information in the simulation environment is ideal, which is different from the real sensor's perception of the physical environment; and the detected targets in the simulation system are limited and cannot Complete simulation of the real world. For example, in the real world, fences of different heights and materials have d...

AI Technical Summary

Problems solved by technology

[0004] Self-driving cars can also be tested on the field, but the scene that the field test can simulate is single and it is inconvenient to collect effective data, making it difficult to quantify and reproduce the test process, and it is impossible to track occasional problems; moreover, the field test is limited to safety requirements. It can only operate at low speed conditions, and it is difficult to carry out tests for extreme conditions, so limited field tests are more suitable for evaluating sensor performance and actuator response

Images