A Shared Control Design for Steering Assistance System Considering Driver Behaviors

Abstract

With the rapid development of advanced driver assistance systems, driver-automation cooperative control becomes more challenging due to the uncertainty of driver behaviors. To this end, this paper proposes a shared control approach to assist drivers in path-tracking and collision avoidance. A high-fidelity CarSim vehicle model embedded driver simulator is built to collect the driver data, based on which the normalized steering angle input and the time to collision are defined to quantify the driving performance. Then, these performance indices are employed to develop the fuzzy controller and determine the control authority for the steering assistance system (SAS). A multi-constraints model predictive control is designed to follow the desired path, and simultaneously ensure the vehicle stability. Different from the traditional constraints handling method, the state constraint is transformed into the input constraint considering the nonlinearity. The effectiveness of the proposed method is verified with two less experienced drivers. The test results show that the path-tracking performance and vehicle stability can be guaranteed with less control effort.