Reactive Obstacle Avoidance for Mobile Robots Using Artificial Potential Fields

Abstract

A mobile robot moving through an environment must avoid the obstacles itencounters, and while a path can be planned in advance when theenvironment is known, a robot often needs to react to obstacles as it sensesthem, steering around them on the fly; the artificial potential field methodprovides a simple and elegant means of doing so. This study examines reactive obstacle avoidance using artificial potential fields and the well-known difficulty of local minima that attends it. The method treats the goal as exerting an attractive pull on the robot and each obstacle as exerting arepulsive push, the robot moving at each instant in the direction of thecombined force, and a robot was navigated from a start to a goal amongobstacles under this scheme. The robot reached the goal by a smooth path that bent away from each obstacle as it approached, the repulsive push from the obstacles steering it clear while the attractive pull drew it onward, the whole behaviour emerging simply from the sum of the forces without any path being planned in advance. The method was computationally light and reacted naturally to the obstacles, which makes it attractive for real-time avoidance, but it is known to suffer from local minima, configurations where the attractive and repulsive forces cancel and the robot halts short of the goal, for instance directly before a large obstacle or in a concave one. The study shows that artificial potential fields provide effective and lightweight reactive obstacle avoidance, well suited to steering around obstacles in real time, while cautioning that the local-minimum problem must be addressed for reliable navigation, and it discusses the forces involved and the means of escaping local minima. KEYWORDS: Obstacle avoidance, Artificial potential fields, Reactivenavigation, Mobile robots, Local minima, Attractive force, Repulsive force,Real-time navigation, Autonomous robots