Modeling and control of a large-span redundant surface constrained cable robot with a vision sensor on the platform

Abstract

A type of cable driven redundant parallelmanipulator is considered that is spanned across a surface, instead of suspended in space. Acontrol strategy needs to be implemented that guarantees positive cable tensions and provides good performance in trajectory tracking. Therefore a workspace control method is designed based on a model of the robot,which includes a solution for the cable tension distribution. As it is hard tomeasure the absolute position of the end-effector, a newapproach is presented where feature detection is used to update an approximated position. Experiments are carried out to compare performances with and without the addition of a vision system for a simple line motion. The error between the platform and the reference position is measured for both cases as well as the cable tensions. Mean absolute errors of 8.8 and 13mm are obtained for the case with and without vision system respectively and positive cable tensions are measured for the complete motion. This indicates that the principle of including feature detection in the position measurement is a suitable control strategy.