An Adaptive SOM Neural Network Method for Distributed Formation Control of a Group of AUVs

An adaptive self-organizing map (SOM) neural network method is proposed for distributed formation control of a group of autonomous underwater vehicles (AUVs). This method controls the AUVs holding their positions in the formation when the formation moves as a whole. The group of AUVs can reach the desired locations in an expected formation shape along preplanned trajectories. The proposed control law is distributed in the sense that the controller of each AUV only uses its own information and limited information of its neighboring AUVs. Formation-control strategies based on self-organizing competitive calculations are carried out with workload balance taken into consideration, so that a group of AUVs can reach the desired locations on the premise of workload balance and energy sufficiency. Moreover, the formation can avoid obstacles and change its shape as needed. The formation is in a distributed leader–follower-like structure, but there is no need to designate the leader and the followers explicitly. All the AUVs in the formation are treated equal to be the leader and the followers, so that important characteristics such as adaption and fault tolerance are achieved. Comparison results with traditional methods and experiments demonstrate the effectiveness of the proposed method.