Motion Planning for Autonomous Underwater Vehicles in Near-Seafloor Operations Using State Lattices

This paper presents a motion planning algorithm to enable planning of safe and efficient motions for autonomous underwater vehicles (AUVs) at low altitudes using state lattices. The algorithm was developed for and tested on a HUGIN 1000 AUV conducting data gathering for optical identification of objects on the seafloor. As the risk to the AUV is greatest at low altitudes, a resolution-complete state lattice A* motion planner was employed to find collision-free resolution-optimal motions. The state lattice A* algorithm was also proven to have a fixed worst-case execution time for a specific motion planning problem. The planner was experimentally verified in a sea trial where a HUGIN 1000 AUV conducted a bathymetric mapping survey of a target area and then planned and traversed a collision-free motion between an operator specified start and goal point. As the HUGIN AUV currently operates using waypoint guidance, the motion was reduced to a sequence of waypoints, resulting in some deviation from the original motion. The route was however traversed collision free by the AUV at an altitude of 10 m. The main contribution of this paper is the adaptation of the state lattice A* planner to a dynamic model of an AUV and validation of the algorithm experimentally.