Initial pose self-calibration of redundant cable-driven parallel robots considering anchor point errors

Abstract

Separated cable-driven parallel robots are widely deployed in large-scale and high-dynamic applications due to their flexible configurations. However, structural parameter deviations, specifically anchor point shifts caused by prolonged operation or equipment reconfiguration, often compromise initial pose accuracy. To address this, this paper proposes a rapid initial pose calibration method. By utilizing incremental cable length changes under controlled platform perturbations, a nonlinear mapping model is formulated to resolve unknown anchor positions and platform poses. The method integrates the Improved Snow Ablation Optimizer with the Levenberg–Marquardt algorithm, employing a rolling iteration mechanism to simultaneously identify anchor point errors and the initial pose. Validation experiments on a custom-built prototype demonstrate a mean positioning error of 1.2 mm under static conditions. These results confirm the method’s high calibration accuracy, ensuring the geometric precision required for subsequent high-dynamic operations.