A random local-DRM path planning algorithm for dual manipulator mobile robots in changing environments

Path planning for mobile robots with high degree of freedom (DOF) is an extreme challenge. Although lots of algorithms have focused on the planning of fixed manipulators and mobile robots with low degree of freedom, seldom of them can be employed to deal with high DOF mobile agents. The unpredictable obstacles and too many freedoms increased computational complexity dramatically. In this paper, a novel and general real-time approach is introduced to solve this problem. The core of this approach can be divided into two phases. By locally employing Dynamic Roadmap Mapping, a mobile robot can seek a collision-free path locally without too much collision detection. And a hierarchy sampling strategy is employed to treat narrow passages. Based on the idea of Rapidly-Exploring Random Tree, a high-level guide is developed by randomly generating subgoals. These two phases collaborate to help generate a desired path. Experimental results show that our algorithm can find out a collision free path in real time for mobile robots with 15 DOF in complex environments in the presence of both stationary and changing obstacles.

I. INTRODUCTION

Since the presence of randomized algorithms, such as the popular probabilistic roadmap method (PRM) [1] and rapidly-exploring randomized tree method (RRT) [2], path planning study has improved significantly. The derivatives of these algorithms can not only solve the traditional piano mover's problems, but also be competent for path planning of simple robots in changing environments. The work of [3] [4] can successfully plan a path for autonomous cars or mobile robots with low degree of freedom (DOF). More recent works [5] [6] also present efficient path planning algorithms in changing environments. Nevertheless, these algorithms can only plan paths for low-DOF mobile agents. References [7] [8] can effectively solve the problems of manipulators or robots fixed on a certain position. They can find a collision-free path with both stationary obstacles and changing ones in environments. However, problems become much more complicated if these manipulators are mounted on a mobile base.