Computational issues in motion planning for autonomous underwater vehicles with manipulators

Abstract

Motion planning with a high number of degrees of freedom (DOF) is computationally demanding. Coupled AUV-manipulators are an example where there can be as many as 14-DOF in a typical dual-arm configuration attached to an underwater vehicle exhibiting 6-DOF motion. In an underwater environment, autonomous vehicles are required to plan motion online, a non-trivial task for coupled AUV-manipulators. This paper examines the computational requirements of high DOF systems and presents a technique which allows fast execution of motion plans. This technique is implemented using distributed search in a 'local' motion planning context. Distributed search is achieved through the execution of a number of subsearches in parallel, where each subsearch contains a unique subset of the DOFs in the system. It is shown how the use of this technique increases the effectiveness of the local motion planning methodology, allowing fast execution of motion plans for systems with a high number of degrees of freedom.