Model Predictive Control of a Robot Driven Vehicle for Testing of Advanced Driver Assist Systems

Abstract

Advanced Driver Assist Systems (ADAS) are becoming more prevalent and more sophisticated in passenger vehicles, with features such as automatic lane keeping, pedestrian detection, and emergency breaking. In line with the increased production deployment of ADAS, testing of these systems are becoming more rigorous with more scenarios needing to be considered every year, see, for example, the ADAS testing conducted by Euro NCAP. To fit the need of placing test vehicles and environment factors in very specific and repeatable scenarios, physical driving robots are commonly used. While most current tests set up the test vehicle in near steady state conditions, e.g., constant speed and straight, in the future, more complex, possibly dynamic scenarios will need to be tested. This paper presents a Model Predictive Control (MPC) strategy for controlling a vehicle along dynamic paths and is easily deployable across different vehicles. Experiment results demonstrating the controller capability for both an electric and a conventional vehicle is presented.