A personalized shared control method for co-driving intelligent vehicles: Considering collision severity and driver capability

Abstract

Human-machine shared control represents a pivotal transitional phase between traditional and fully autonomous driving, and has increasingly attracted scholarly attention in recent years. To address the issue of human-machine conflict in shared steering control, this study proposes a personalized human-machine shared control approach that accounts for individual driver differences. First, to characterize the interaction between the driver and the Advanced Driver Assistance System (ADAS), a non-cooperative game-based human-machine shared control framework based on Model Predictive Control (MPC) is established. Subsequently, to achieve personalized shared control, a driving authority allocation strategy is developed based on assessing collision severity and the driver capability, enabling individualized shared control through personalized authority weights. Ultimately, the effectiveness of the proposed non-cooperative Nash game-based human-machine shared steering strategy and the flexible authority allocation scheme is validated through driver-in-the-loop experiments. Experimental results show that the proposed human-machine shared control strategy mitigates the driver’s operational burden, improves driving safety, and enables personalized authority adaptation in accordance with individual driver characteristics.