Optimum Design of Serial Robots

Abstract

An optimum design of an industrial robot can be achieved from different point of views. For example, a robot can be conceived from the standpoint achieving maximum workspace or minimum weight, etc. In this paper, the objective is to arrive at a robot design that will require optimum driving torques/forces at its joints to perform tasks within its workspace. Such a design will automatically save energy. Note that these torques/forces at the joints are highly dependent on the mass and the inertia properties of the robot’s links. Therefore, these quantities were minimized by determining the optimum masses and optimum mass centers and finding out the corresponding inertia properties of the moving links. Such an approach was briefly introduced earlier by the authors with the help of a simple two-link planar arm. In this paper, the concept is generalized and demonstrated with the help of a complex robot, a 6-degrees-of-freedom PUMA robot. To achieve the design for optimum driving torques/forces at the joints, the concept of equimomental system of point masses was introduced, which helped to obtain the optimum locations of the mass centers of each link quite conveniently. However, to compute the driving torques/forces recursively for such equivalent point mass systems, the decoupled natural orthogonal complement matrices for point masses (DeNOC-P) was derived. It has led to a simplified algorithm for obtaining driving torques/forces. The proposed algorithm for optimization is illustrated with the help of a PUMA robot.