Fast underwater vehicle manipulator system motion planning in GPUs

In this paper, we present a method for global motion planning of underwater vehicle manipulator systems (UVMS). The algorithm can derive a feasible path for UVMS equipped with an n-degree of freedom manipulator for performing autonomous underwater operations. The environment of the system is represented by a 4D Bump-surface that is used for obstacle avoidance. To improve the performance of the proposed method, the parallel calculation of the obstacle avoidance algorithm is considered, leading to a 46-230 times faster implementation using a Graphics Processing Unit. A path following task in an environment cluttered with obstacles is considered for the validation of the method, while a UVMS with a mounted 6-dof manipulator performs the task. Both parallel and serial approaches are evaluated and discussed.