Optimal Obstacle Avoidance Path Planning for Autonomous Underwater Vehicles

A reliable obstacle avoidance capability is a critical link in the quest for true underwater vehicle autonomy. Investigations in this area have culminated in a three-dimensional obstacle avoidance algorithm for autonomous underwater vehicles (AUVs). Using this algorithm, an AUV outfitted with an obstacle avoidance sonar (OAS) and a precise navigation system can localize obstacle positions and safely navigate a dense, unknown obstacle field. The algorithm overcomes the physical constraints of the OAS such as narrow field of view, missed detections, false alarms and poor angular resolution. Furthermore, based on the algorithm's present knowledge of obstacle positions, dynamic programming techniques plan the optimum safe path through the obstacle field. Additional logic ensures that in spite of the AUV's restricted maneuverability and the OAS's limited coverage, the vehicle always maintains the minimum, safe standoff distance from all obstacles. Computer simulation has verified the obstacle avoidance algorithm's integrity in six dense, box canyon obstacle fields and two random obstacle fields. In all cases the algorithm safely guides the AUV through the obstacle field and localizes many of the obstacle positions.