Efficient computation of configuration space transforms for collision-free motion planning

Abstract

Robot motion planning in workspace suffers from issues like multiple Inverse Kinematics solutions and computationally expensive collision detection. The concept of Configuration Space was thus introduced in robotics to ease the problem of motion planning. This paper presents an approach to transform obstacles in the workspace to obstacles in the Configuration space for a 2 Degree of Freedom (DOF) R-R planar manipulator. The proposed algorithm uses a single primitive using which various convex and non-convex obstacles in the world can be transformed to their configuration space equivalent. The explicit construction of entire obstacle in configuration space is computationally expensive. Thus, only the boundaries of obstacle in configuration space have been generated by utilizing the properties of configuration space transforms. This approach can be generalized, with some modifications, to represent configuration space obstacles for robots with higher DOF. After reading this paper, the reader should be able to understand the need for configuration space in robotics, some of its essential properties and be able to construct one for planar manipulators.