This paper develops an indoor rapid testing platform for Autonomous Vehicles (AVs) based on vehicle-in-the-loop simulation. This platform comprised of four subsystems, which respectively are: (1) A drum-based mechanical test bench; (2) A compute-aided measuremental control system; (3) A software-based scenario and sensors simulated system; (4) An automatic evaluation system for the testing result. The proposed platform couples the AV being tested, the virtual simulation scenarios and the test bench to construct a closed-loop system. In the mechanical test bench, the driving resistance is simulated by independently loading the torque on the rollers, which support four wheels of the AV, so that the different road adhesion coefficients can be simulated. The mechanical test bench has 3 degrees of freedoms (DoFs), and can rotate with pitch, roll and yaw angles, which can simulate the geometry of roads. Furthermore, we adopted virtual reality technology to simulate various road traffic scenarios for verifying the functions like intelligent perception and behavior decision-making of the AV, by the flexible integration of vehicle dynamics model, sensor simulators, simulations of complex traffic environments and test cases. Lastly, to verify the effectiveness of the test platform, this paper selects lane change of an AV as the use case, many experiments were conducted on the proposed platform. The experimental results show that the tested AV can efficiently execute the lane-changing maneuver under various conditions on the proposed platform. Compared with actual road test, the test efficiency of the platform has been greatly improved.

CICTP