Motion-planning for welding robots

Abstract

We present an approach to tackle general motion planning problems in high dimensional configuration spaces. The position of a robot in a workspace can be uniquely described by a configuration vector. This vector for example encodes the angles between consecutive links of a robot manipulator, the length of this vector is the number of degrees of freedom (dof) of the robot and the dimension of the configuration space. For a scene of obstacles in the workspace the motion planning problem aims to find a collision-free motion between two given configurations of the robot. The running time of all exact algorithms is exponential in the number f of degrees of freedom in the worst case. This is due to the fact that the complexity of free space can be /spl Omega/(n/sup f/), where n is the number of constraints. Obviously these approaches are not well suited for practical applications, where f is usually larger than three. Here, we put forward the claim that in practical situations it may not be necessary to compute an optimal path. We develop a new heuristic taking advantage of the "sparsity" of the obstacles, which means that the clearance of the motion we are seeking is large. We focus especially on welding robots to argue that our assumptions are practical. Our algorithm is based on the A*-algorithm and pure clearance computations in the work space. Our algorithm has been implemented in its general form to verify and test its performance, that means there are no restrictions of the dimension, the type of the robot etc. The efficiency of our method is confirmed by experiments in simulated environments.