Pose Error Real-Time Prediction and Compensation of a 5-DOF Hybrid Robot Based on Laser Tracker and Externally Mounted Encoders

Error compensation is an effective approach for robots to improve accuracy. This paper presents a novel method to predict and compensate for pose error of a 5-DOF hybrid robot on-line with the usage of externally mounted encoders, concentrating particularly on compensating dynamic errors on the account of changes in external forces or disturbances. A novel method to estimate pose error is proposed employing the offline sampling data from a laser tracker as well as the online measurement data from the external encoders. A real-time procedure for pose error prediction and compensation is applied into the NC system, which involves two successive steps: (1) calculation of pose error based on the online measurement from externally mounted encoders and the offline data measured by the laser tracker employing the moving least squares algorithm, and (2) compensation for the command pose in every interpolation cycle. Experimental verification shows that the residual inaccuracy of pose error prediction is reduced by 61% with respect to that only estimated from offline data under the condition of changing loads and the deviations of predicted errors respect to actual errors are within 5% under a constant load.