Decentralized Collision Avoidance via Fuzzy Potential Fields

An intelligent and cooperative collision avoidance method is proposed based on a combination of enhanced potential fields and fuzzy inference systems (FIS), where a genetic algorithm is used to optimize the FISs. The proposed approach provides a near-optimal and collision-free path in an environment with static and dynamic obstacles taking into consideration potential uncertainties. Furthermore, it is able to resolve local minima and goal non-reachable with obstacles nearby issues that exist in the traditional artificial potential field approach with minimum computational burden. A simple scenario modeled based on the issues faced by UAVs in such environments is used for training the system, and a complex scenario containing a number of dynamic UAVs in the presence of static obstacles is considered to validate the performance of the proposed approach.