Game-Theoretic Human-Machine Shared Steering Control Strategy Under Extreme Conditions

Human-machine shared control is a necessary transitional stage from manual to fully autonomous driving. A novel game-theoretic shared control strategy is proposed to ensure driving safety under extreme conditions. Specifically, a differential game framework is employed to mathematically model the dynamic human-machine interaction. The shared controller is developed based on the nonlinear vehicle dynamics and model predictive control (MPC). The piecewise affine (PWA) approach is utilized to linearly approximate the vehicle system and derive the Nash equilibrium solution of the game problem. Furthermore, a linear weighted shared driving strategy is designed to dynamically allocate the driving weight based on the level of driving risk and human-machine conflict. Simulation results in extreme conditions verify that the shared control strategy can fully exploit tire force to alleviate the driver's burden and compensate for the driver's misoperation, improving lateral stability.