Abstract:

In this work, focusing on self-driving vehicle safety testing in a complex environment, a mixed-reality-based self-driving vehicle testing method is proposed. This method maps the virtual scene in the simulation environment to the real test environment. Then, performance indexes such as intelligent perception and the behavioral decision of the self-driving test vehicle are rapidly estimated using a sensor and data fusion to establish a closed-loop test system. This system involves the mixed-reality test scene, control center, and test vehicle. Finally, based on the verification of a case study, the proposed method is found to meet the test requirements of self-driving vehicles in various scenarios.

Key words: mixed reality, self-driving, test methods, test scene, background vehicle, control center, vehicle-in-the-loop