AI-based calibration by using a motion capture system for Autonomous Mobile Robots

This paper demonstrates the feasibility of an AI-based calibration method for enhancing the positioning accuracy of Autonomous Mobile Robots (AMR) by using a motion capture system. While similar calibrations have been tested on other robot platforms, this work explores their applicability to AMRs. AMRs play a critical role in material transport between workstations, necessitating precise positioning at transfer stations to minimize errors and maximize process stability. By employing a highly accurate external measurement system, specifically an optical motion capture setup, this study aims to automatically calibrate the AMR's positioning system. Previous calibration methods remain largely manual with the need of human interaction in the calibration process. This approach aims to automate the calibration by reducing the needed interactions. To achieve this a machine learning algorithm is trained on the discrepancies between the actual and intended positions, allowing for real-time adjustments and improved accuracy. Experimental results demonstrate a 20% improvement in positional accuracy, demonstrating that the approach is viable for AMRs.