AUV Guidance System for Dynamic Trajectory Generation

This paper considers an AUV guidance system for collision-free transit and complete area coverage. A topologically organized neural network model is used to represent the operating environment. The activity of each neuron, characterized by a shunting equation, is used to represent the local environmental information. Targeted areas have the highest values. The AUV moves from areas with low activity to areas with higher activity like in a potential field navigation. The kinetic constraints of the AUV are taken into account by using Dubins theory to generate feasible paths.