AUV Path Planning for Target Search Using Bioinspired Neural Network

Abstract

The bio-inspired neural network is utilized and improved to solve the autonomous underwater vehicle (AUV) path planning problem for target search mission in ocean environment. First, the standard model of Glasius bio-inspired neural network (GBNN) is constructed to reflect the characteristics of the searching region. The activity value of each neuron is updated by the value propagation among neighbors, as well as the stimulus of itself from the detection reward and the constraint of obstacle avoidance. On the basis of activity distribution, the AUV’ s next waypoint is chosen greedily. Second, the standard GBNN is modified to further enhance the searching efficiency. The connection weight between neighbor neurons is modified by the voyage time considering ocean current. The future reward attracted from all the areas with peak probability is introduced into the neuron activity. In addition, when multiple AUVs perform the search mission, the whole region can be divided into parts with the same size by the Voronoi diagrams. The simulation results demonstrate the high efficiency and strong robustness of our method.