Obstacle avoidance using complex vector fields

Abstract

Obstacle avoidance is an important part of an autonomous navigation system. One method of obstacle avoidance is the use of simple vector fields. Simple vector fields used for obstacle avoidance consist of a repulsive force centred on the obstacle (obstacle vector field) that guides the vehicle away from the obstacle. This is added directly to an existing vector field (environmental vector field) that guides the vehicle towards the goal to produce a resultant force on a vehicle. This method has been shown to avoid obstacles but does have limitations, the most common of which is that the vehicle can become trapped in a ‘U’ shaped dead end. This study outlines a method of vector field navigation that blends the environmental and obstacle vector fields instead of adding them directly together. The navigation method will also use rotational vector fields for obstacle avoidance instead of repulsive forces. It will be shown that this method can be used for avoiding obstacles as well as evading the ‘U’ shaped or dead end trap.