This thesis presents a method for compensating position errors on a rail guided shuttle system. The shuttle is equipped with an absolute position sensor that measures the true position in 0.5 to 0.6 𝑚 intervals in addition to a motor encoder. The proposed algorithm is an extension to the recursive least squares method and learns to estimate the position error based on the discrepancy between absolute position sensor and motor encoder. Additionally, the velocity and acceleration obtained using a Kalman filter as well as the motor current are used as inputs. The algorithm is based on a two- or threeparameter model that was developed using a data driven approach. Two parameters have been identified to be the main contributors to accurate position error prediction. A velocity-proportional term captures wheel radius inaccuracies and a square root velocity times motor current term captures friction effects between wheel and rail. The main goal is to replace the existing positioning system that relies on an external encoder, which is less affected by positioning errors compared to the motor encoder. The algorithm was implemented and validated on a Maxon motor controller under realworld conditions. It has been shown to outperform the existing system by reducing the average positioning error from 3.27 𝑚𝑚 to 2.20 𝑚𝑚, which corresponds to an increase in positioning accuracy of 33%.
Adaptive Position Error Compensation for a Rail-Guided Shuttle System
Bauernfeind, M. P. C. (Author). 2025
Student thesis: Master's Thesis