Fast Kinematic Re-Calibration for Industrial Robot Arms

Area of Science:

• Robotics
• Mechatronics
• Control Systems

Background:

• Accurate kinematic modeling is crucial for safe robotic operations.
• Manufacturing errors, especially joint offsets, cause significant position inaccuracy in robots.
• Existing calibration methods often require external devices or fixtures.

Purpose of the Study:

• To develop a fast robot recalibration approach using factory-calibrated controllers as an 'oracle'.
• To extract calibrated intrinsic robot parameters not directly available.
• To minimize kinematic mismatch by compensating for joint zero position error and link length variations.

Main Methods:

• Utilized the factory-calibrated controller as a data source for recalibration.
• Developed a method to compensate for joint zero position errors and link length variations.
• Applied the method to a Kinova Gen3 ultra-lightweight robot.

Main Results:

• Successfully minimized the kinematic mismatch between ideal and factory-calibrated robot models.
• Experimental analysis validated the proposed recalibration method.
• Demonstrated significant error reduction compared to uncalibrated models.

Conclusions:

• The proposed fast-recalibration approach effectively improves robot kinematic accuracy.
• Leveraging factory calibration data offers a practical solution for enhancing robot performance.
• The method is validated experimentally for real-world robotic applications.