A Redundancy-Aware Error Model for Kinematic Calibration of Redundantly Actuated Overconstrained Parallel Robots

Abstract

For redundantly actuated overconstrained parallel robots (ROPRs), the existing general error models for kinematic calibration often omit their redundantly actuated and overconstrained characteristics, affecting the precision of the calibration. To address this issue, a general redundancy-aware error model (RAEM) is proposed. First, the configuration space of the ROPRs is implicitly constructed to expose the constraints on both the active joints and the geometric errors resulting from these two characteristics. To mitigate the consequent effects, the restrained deviations of redundant active joints are portrayed both in the RAEM and in the forward kinematics during each calibration iteration. A feasible error space is, thus, formulated to impose geometric error constraints on the identified errors. The introduced method can theoretically prevent the violation of configuration constraints. Finally, the superiority of the proposed RAEM is validated by kinematic calibration simulations and experiments. The results show that the proposed RAEM achieves better calibration precision than existing methods.