Human-Vehicle Cooperation Approach Based on Seamless Online Authority Transfer and Robust Control

Human-vehicle cooperation approach enables a human operator and an automation system to collaboratively handle a common task. This method poses numerous challenges due to human-machine transfer authority, especially lanekeeping applications. This paper introduces a design for a lanekeeping assistance system based on human-vehicle cooperation. Our proposed method assists the human in driving efficiency through the implementation of seamless online authority transfer and robust control. To calculate the level of collaboration between the human driver and the automation, we employ a radial basis function neural network trained on data including lateral jerk, steering angle error, and steering angle derivative. Subsequently, the sliding mode control algorithm computes the steering angle to ensure precise vehicle lanekeeping. Validation through numerical simulations on curved trajectories demonstrates the effectiveness of our strategy.