Deadlock avoidance for robot navigation with potential field method in an unknown environment

We propose an efficient obstacle and deadlock avoidance strategy for mobile robots in an unknown environment using a potential field method. A well-known downside of potential field methods is their tendency to get stuck in deadlock positions, originating because of the balancing between the attractive force to the destination and the repulsive forces from the obstacle. To account for this shortcoming, we modify the way the robot receives repulsion forces by introducing virtual constraints that fill in hollow obstacles, and design a strategy for generating temporary intermediate destinations. Numerical simulations confirm the effectiveness of the proposed methodology.