Precision Robotic Assembly Using Attractive Regions

Abstract

This paper describes a method for peg-in-hole insertion based on attractive regions a new approach to decompose 6D configuration space into two 3D subspaces. An impedance controller, in combination with the attractive regions, is integrated in a novel controller whereby human-like assembly performance is achieved without using a force sensor. In this paper, the insertion task uses three different pegs, as shown in Fig. The strategy is to decompose the 6 DOF space in which the peg moves into two low-dimensional subspaces. The controller uses stereo-triangulation and a k nearest-neighbor (k-NN) algorithm to identify the parts. An impedance controller monitors the forces (estimated by motor current rather than a force transducer) applied to the system and uses events rather than time to determine the state and the evolution of the assembly process. The approach is experimentally verified using a Barrett Technology's WAM Arm, a backdrivable 7 DOF robotic manipulator.