Sensor-Free Self-Calibration for Collaborative Robots Using Tri-Sphere End-Effector Toward High Orientation Accuracy

Abstract

Collaborative robots often exhibit limited absolute accuracy despite high repeatability, necessitating cost-effective calibration solutions. This paper presents a novel sensor-free self-calibration method for collaborative robots using position and distance constraints. A tri-sphere end-effector with precision balls and magnetic holders enables repeatable Tool Center Point (TCP) positioning (<0.01mm) through hand-guiding, where the three-sphere configuration crucially enhances the orientation calibration accuracy compared to a single-sphere approach. The proposed device eliminates expensive external sensors while establishing geometric constraints through workspace-wide TCP engagements. By analyzing relative position/distance errors between multiple configurations, the method identifies kinematic parameters via a Local Product of Exponential (Local POE) based error model. Experiments demonstrated a 91.7% position error reduction (7.98mm to 0.66mm) and 69.6% orientation improvement (0.0069rad to 0.0021rad), achieving comparable accuracy to laser-tracker methods at <1% device cost. This approach offers a low-cost, mechanically robust solution for enhancing collaborative robot accuracy in industrial applications.