GETTING EXACT INFORMATION FROM THE INVERSE JACOBIAN MATRIX OF PARALLEL AND SERIAL ROBOTS

摘要

The inverse jacobian matrix of a robot contains highly useful information about the performances of the robot, for example about its accuracy and dexterity. Unfortunately these information are available only mostly through the eigenvalues of this matrix which are pose and design parameters dependent. Interval analysis based algorithms is presented that allow to extract useful information from any matrix such as minimal and maximal eigenvalues and minimal condition number.