Journal of Intelligent and Robotic Systems 19: 73–88, 1997. 73 c  1997 Kluwer Academic Publishers. Printed in the Netherlands. A Robust Method for the Concurrent Motion Planning of Multi-Manipulators Systems R. V. MAYORGA and A. K. C. WONG Department of Systems Design Engineering, University of Waterloo, Waterloo, Ontario, Canada. e-mail: rene@watnow.uwaterloo.ca Abstract. In this article a robust and simple procedure for the on-line concurrent motion planning of multi-manipulators is presented. The approach is based on solving for each manipulator a linear system of equations taking into account a vector for motion planning, and an original scheme for the appropriate perturbation of the pseudoinverse matrix. This method can pursue simultaneously both motion coordination and singularities prevention in real time in a sensor based environment. These properties make it suitable for fully autonomous or telerobotic systems operations. Key words: concurrent motion planning, multi-manipulator systems, redundancy resolution 1. Introduction Currently many industrial tasks are performed efficiently with the use of dual or multi-manipulators systems. In general, these systems execute mainly two types of operations: one in which all manipulators cooperate for a given task, such as carrying an object; and another type in which each manipulator accomplishes its own task independently in a common workspace. The first kind has been widely studied in the literature [26, 5], mainly from a control problem point of view, [5]. Alternatively, the motion planning complexity of several manipulators and objects moving in a common workspace [3] has limited the treatment of the second type of actions to relatively few studies on point to point tasks for particular nonredundant manipulators [7, 22], in a nontruly concurrent fashion [23, 8, 5] and mainly to the development of off-line approaches. In this article a novel, robust and simple procedure for the on-line concur- rent motion planning of multi-manipulators systems is presented. The approach is based on, conceptually considering redundant manipulators, formulating for each manipulator an inverse kinematics problem under an inexact context. This approach allows to deal with the motion planning utilizing an easily computable vector to be utilized in a null space; whereas the avoidance of singularities is attained using a simple scheme for the proper pseudoinverse perturbation. The inverse kinematics problem for redundant manipulators is usually studied under an exact context by global [20, 17] or local approaches [14]. The glob- al approaches have the property that the solution is searched for the entire task space at once, whereas the local approaches compute a solution at each instant of