Investigation of Time Efficiency and Near Singular Behavior of Numerical Robot Kinematics

摘要

This paper introduces several modifications to the numerical methods used to find inverse kinematics solutions of manipulators. Modified Newton-Raphson and modified predictor corrector algorithms are discussed and compared. These modifications contribute significantly to the reliability, stability, and time efficiency of these algorithms. These modified algorithms always converge if a physically reliable robot configuration exits, and are able to numerically determine (or approach) singular positions of the robot with desired numerical precision. The modified predictor corrector based inverse kinematics is much faster than the one based upon the modified Newton Raphson algorithm.