Neural Adaptive Control of a Robot Joint Using Secondary Encoders

Jonas Weigand, Magnus Volkmann, Martin Ruskowski

In: Karsten Berns , Daniel Görges (Hrsg.). Proceedings of the 28th International Conference on Robotics in Alpe-Adria-Danube Region (RAAD 2019). International Conference on Robotics in Alpe-Adria-Danube Region (RAAD-2019) June 19-21 Kaiserslautern Germany Seiten 153-161 Advances in Intelligent Systems and Computing (AISC) 980 ISBN 978-3-030-19648-6 Springer International Publishing 2019.


This paper aims to reduce gearbox errors on industrial robots with a feed forward, neural adaptive control. The algorithm combines two networks, one for control and one for system identification. In order to achieve a high precision and generality on untrained data, a Runge-Kutta Neural Network is used for black-box identification of a nonlinear robot joint. Secondary encoders as additional angle sensors measure the gearbox error and are used for supervised learning. The presented algorithm is capable of online application and reduces gearbox errors in a nonlinear simulation.

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence