Control of human-machine cooperative cart by region attractor to achieve both obstacle avoidance and trajectory tracking

Abstract

In this paper, we develop a semi-autonomous steering system in which the cart moves toward the goal along the reference trajectory in region avoiding obstacles with human operation. First, the passing points are used as a design variable to modify the reference trajectory away from the wall and make steering angles in the limitation. Next, many parallel trajectories are generated considering the curvature radius of the reference trajectory. Then, a region attractor is designed by a vector field converging to parallel trajectories with human operation. The reference trajectory in the region is switched by human manipulation, and a controller that determines the steering angle based on the position and posture information of the cart is designed. Finally, experiments are conducted to verify that the cart reaches the goal avoiding obstacles.