Control of Cable-Driven Parallel Robots Using Uncalibrated Visual Servoing

This paper presents an implementation of Uncalibrated Visual Servoing (UVS) for controlling a Cable-Driven Parallel Robot (CDPR), eliminating the need for prior calibration or explicit kinematic modeling. Traditional CDPR control methods rely on precise calibration, which is often impractical in dynamic or unstructured environments. To address this, we employ UVS with real-time visual feedback using an eye-in-hand camera configuration and ArUco marker tracking to estimate a global static image Jacobian. We evaluate three control strategies: Traditional Kinematic Control, Single-Point UVS, and Three-Point UVS; assessing their accuracy in aligning with static visual targets. Results show that while Kinematic Control achieves lower error rates, Three-Point UVS offers improved adaptability and platform orientation, indicating UVS's potential for rapid deployment and low-cost, calibration-free control in real-world CDPR applications. Future work will explore adaptive Jacobian estimation and multi-point tracking in dynamic environments.