Error Detection and Prediction Algorithms: Application in Robotics

This article presents research on error detection and prediction algorithms in robotics. Errors, defined as either agent error or Co-net error, are analyzed and compared. Three new error detection and prediction algorithms (EDPAs) are then developed, and validated by detecting and predicting errors in typical pick and place motions of an Adept Cobra 800 robot. A laser Doppler displacement meter (LDDMTM) MCV-500 is used to measure the position of robot gripper in 105 experiment runs. Results show that combined EDPAs are preferred to detect and predict displacement errors in sequential robot motions.