Parallelized 3D Inverse Kinematics with Multiple Objectives

A strategy of parallel computing (HPC) is proposed to solve the problem of inverse kinematics in 3D with multiple objectives using the damped least squares method, also known as the Levenberg-Marquardt method. The program solves the problem of moving one or more end-effectors of a robot to a desired target position by manipulating its joints, which are the degrees of freedom. The HPC strategy consists in parallelizing the motion calculations required to accomplish the different objectives. Tests were conducted with a simulation of a ping-pong game using multiple balls. Robots are placed at each end of a table, the movement of the balls is predicted integrating its position numerically, and the robots' end-effectors are moved to hit the balls. The amount of end-effectors correspond to the number of balls, with priorities assigned to each of them.