An on-line relative position and orientation error calibration methodology for workcell robot operations

Abstract

Relative position and orientation inaccuracy always exists between a robot and the equipment with which it operates, especially in batch-type production cells that are subjected to dynamic changes. This inaccuracy causes robot relative positioning errors, and may even result in operation failure if the off-line programmed moving path is implemented without adjustment. To make use of off-line programming and simulation tools, an on-line calibration methodology for robot relative positioning inaccuracy was developed in this study. This methodology eliminates the need for time-consuming off-line calibrations relying on accurate models and expensive devices. An industrial robot system was enabled to detect and compensate automatically for relative positioning errors by incorporating a vision system, a 3-D force/torque sensor, and control strategies involving neural networks. The experimental results showed that this methodology is valid and robust in calibrating the relative position and orientation errors automatically without the need for mathematical models and complex off-line calibration procedures for model parameters. Consequently, batch-type production cells would be more flexible, adaptable and intelligent in accommodating dynamic workcell changes with less human effort.