Evaluation of hand-eye and robot-world calibration algorithms for TMS application

Area of Science:

• Robotics
• Medical Engineering
• Computer Vision

Background:

• Accurate robot calibration is crucial for precision in medical applications like Transcranial Magnetic Stimulation (TMS).
• Hand-eye and robot-world calibration are essential for integrating robotic systems with medical devices.

Purpose of the Study:

• To compare three calibration methods (QR24, SGO, QUAT) for a TMS system.
• To evaluate the performance of each method based on accuracy and computational time.

Main Methods:

• Implemented and tested non-orthogonal (QR24), stochastic global optimization (SGO), and quaternion-based (QUAT) calibration approaches.
• Utilized a 7 degrees of freedom Panda robot, Polaris Vicra camera, and SofTaxic Optic software for experimental setup.
• Evaluated performance using translation and rotation errors across various datasets and calibration points.

Main Results:

• The SGO method demonstrated superior performance with the lowest translation and rotation errors.
• SGO achieved high stability across different datasets and numbers of calibration points.
• Optimal performance with SGO was observed in a small workspace (0.05m radius) using approximately 150 calibration points, yielding errors of 0.83 ± 0.35mm and 0.22 ± 0.12 degrees.

Conclusions:

• The SGO method is the most effective for hand-eye and robot-world calibration in TMS applications due to its accuracy and stability.
• While SGO has a higher computational time, this is not a limiting factor for TMS systems.
• Achieving optimal results with SGO requires careful consideration of workspace size and the number of calibration points.