A Calibrator Fuzzy Ensemble for Highly-Accurate Robot Arm Calibration

Area of Science:

• Robotics
• Machine Learning
• Manufacturing Automation

Background:

• Industrial robot arm absolute positioning accuracy is crucial for automated manufacturing processes.
• Existing data-driven robot arm calibration methods often use single learning models, leading to accuracy loss due to limited solution space representation.
• Improving calibration accuracy is essential for advanced manufacturing applications like automatic assembly.

Purpose of the Study:

• To propose a novel calibrator fuzzy ensemble (CFE) method to enhance industrial robot arm calibration accuracy.
• To address the limitations of single learning models in existing data-driven calibration approaches.
• To achieve impressively high calibration accuracy for industrial robot arms through an innovative ensemble technique.

Main Methods:

• Implementation of eight diverse data-driven calibrators utilizing sophisticated machine learning algorithms.
• Development of a fuzzy ensemble framework to integrate the outputs of the eight individual calibrators.
• Extensive experimentation on an ABB IRB120 industrial robot using MATLAB for validation.

Main Results:

• The proposed calibrator fuzzy ensemble (CFE) demonstrated superior performance compared to state-of-the-art calibrators.
• CFE achieved a significant reduction in maximum error by 8.59% on an ABB IRB120 industrial robot.
• Individual base models within the ensemble ensured high accuracy, contributing to the overall improved performance.

Conclusions:

• The calibrator fuzzy ensemble (CFE) method offers a robust solution for improving industrial robot arm positioning accuracy.
• CFE effectively overcomes the limitations of single learning models by leveraging the diversity of multiple algorithms.
• The proposed method shows great potential for practical implementation in real-world manufacturing applications.