Model Predictive Obstacle Avoidance Control for Vehicles with Automatic Velocity Suppression using Artificial Potential Field

Abstract In this paper, moving obstacle avoidance in narrow environment is studied. Recently, electric wheelchairs are widely used, which has independent driving wheels modeled as a two-wheeled nonholonomic vehicle model. One of the challenges on obstacle avoidance in real environment is that the obstacle position observed by sensors is usually fluctuated due to noise; it is difficult to pass through the obstacles safely. To overcome this difficulty, we apply time-state control form to divide the system into two subsystems: path dynamics and velocity dynamics so that we can separate the path planning from velocity control, which enables us to transform each subsystem into linear ones. The path tracking is achieved by linear control synthesis design, while running velocity can be tuned depending on the distance to the obstacles. Then we apply model predictive control (MPC) for each subsystem in order to realize obstacle avoidance and optimal velocity control. We impose constraints on MPC representing prohibited area for obstacle avoidance on path following control. In this paper, we show effectiveness of the proposed method by numerical simulations assuming a real electric wheelchair. In the neighborhood of obstacles, the vehicle decelerated in accordance with the distance to the obstacle. It is shown that the effectiveness of the proposed method is supported by numerical simulations.