Shared Control Strategy Based on Driver’s Trajectory Following Intention

Although everything brought by driverless technology is full of attractiveness, until the related technologies and laws are perfected, human-machine shared control is still the best choice at present. There are two main methods in the current shared control strategy. One is that the assisted driving system guides the driver to control the vehicle through voice or tactile. This method cannot guarantee the effectiveness of the assist function when the driver is not attentive. The second is to take over the vehicle based on the risk assessment algorithm by monitoring the vehicle's dynamic state and the surrounding environment. This method can improve the safety of the vehicle, but its comfort is poor. This paper studies the real-time identification method of driver's intention, by constructing the steering angle envelope based on the planned trajectory, establishes a flexible control transfer mechanism, through trajectory switching within the kinematic envelope, the human-machine shared control for trajectory following is realized. Firstly, a series of safety trajectories are obtained by constructing a safe envelope for vehicle kinematics. Then, by using the decoupling characteristics of the steer-by-wire system, the driver's intention recognition algorithm is established by constructing the accuracy envelope of the steering wheel angle, and a trajectory that meets the driver's intention is selected to ensure the driver's confidence. Finally, the path tracking controller using a linear time-varying model prediction algorithm will control the vehicle to follow the trajectory determined by the driver. In this way, unlike driverless technology or some other assisted driving systems, by constructing a flexible transfer mechanism of control rights based on the steering angle envelope, although sacrificing driver expectations within a certain range, not only can the steering wheel shake be filtered to ensure driving comfort, but also the accuracy of trajectory tracking can be improved.