A General Algorithm for Dynamic Control of Multilink Robots

Abstract

One of the major tasks in the development of sophisticated methods for robot control is finding algorithms that are appli cable to a wide range of different robots. Another major task is to find algorithms that can be used for on-line control. Several attempts have been made to solve one or the other of these problems. This article presents a general method for solving both. It can be applied to any redundant or nonredundant robot with rigid subunits connected by revolute joints in a tree-like structure. The number of joints can be chosen arbitrarily large without singularity problems, in contrast to traditional inverse kinematics models. A system based on the presented method needs no modifications when changing from one specifrc robot to another structurally or kinematically completely different robot, only input parameters describing the robot are different. The method is founded on a very general algorithm for finding the Lagrangian equations of motion for a robot subject to a given set of artificial forces. The generality and power of the method are demonstrated by two examples. In the first example the method is used to simulate a 25-degrees-of-freedom (DOF) snake-like robot moving through a maze. In the second example the method is used to control an actual application: an 8-DOF industrial robot system welding a ship section.