Trajectory-Prediction-Based Parallel Shared Steering Control for Human-Machine Cooperative Driving With Obstacle Avoidance

This paper proposes a novel trajectory-prediction-based parallel shared steering control method, for facilitating obstacle avoidance in collaborative human-machine driving scenarios. A driving behavior model of human drivers is modeled using a model predictive control strategy. Based on it, the local trajectory of the driver can be predicted. Then, a 2D Gaussian distribution function is utilized to create a repulsive zone encircling the obstacle vehicles. Moreover, the obtained local predicted trajectory of drivers is utilized for the driving risk assessment, which provides a predictive evaluation of the driving risk of drivers. Finally, fuzzy mapping rules are designed to achieve a smooth transfer of control weight between the driver and the automated system. By employing the suggested shared control approach, greater emphasis is placed on the human driver to enhance driving enjoyment, particularly when the driving conduct poses a lower risk. Otherwise, if the driving behavior poses great driving risks, a larger portion of control authority is shifted from the driver to the automated system to ensure driving safety. The effectiveness of the new approach is showcased via a simulation involving human interaction, illustrating that the shared control strategy is adept at ensuring driving safety consistently. Simultaneously, it successfully maintains the driver's enjoyment and facilitates a seamless transfer of control between the driver and the automated system.