Robust Cable-Length-Differences-Based Position Tracking for a Cable Driven Parallel Robot

Abstract

Developing an affordable and versatile robotic platform is desirable for various industries and applications. Cable driven parallel robots exhibit promising attributes for this objective, but the requirement for accurate end effector position information can necessitate costly additional sensors. This paper describes enhancements to a position tracker along with a neural network-accelerated, faster approximate position initialization utilizing cable length differences. These differences can be inferred from stepper motor movement and the integrated force control system without the need for additional sensors. Our contributions enhance tracking robustness against noise and systematic errors, which is important for low-cost and poorly calibrated hardware in scenarios relevant to the widespread adoption of accessible robotic tools.