An Online Learning Procedure for Feedback Linearization Control without Torque Measurements

M. Capotondi, G. Turrisi, C. Gaz, V. Modugno, G. Oriolo, A. De Luca
Proceedings of the Conference on Robot Learning, PMLR 100:1359-1368, 2020.

Abstract

By exploiting an a-priori estimate of the dynamic model of a manipulator, it is possible to command joint torques which ideally realize a Feedback Linearization (FL) controller. The exact cancellation may nevertheless not be achieved due to model uncertainties and possible errors in the estimation of the dynamic coefficients. In this work, an online learning scheme for control based on FL is presented. By reading joint positions and joint velocities information only (without the use of any torque measurement), we are able to learn those model uncertainties and thus achieve perfect FL control. Simulations results on the popular KUKA LWR iiwa robot are reported to show the quality of the proposed approach.

Cite this Paper

BibTeX
@InProceedings{pmlr-v100-capotondi20a, title = {An Online Learning Procedure for Feedback Linearization Control without Torque Measurements}, author = {Capotondi, M. and Turrisi, G. and Gaz, C. and Modugno, V. and Oriolo, G. and Luca, A. De}, booktitle = {Proceedings of the Conference on Robot Learning}, pages = {1359--1368}, year = {2020}, editor = {Kaelbling, Leslie Pack and Kragic, Danica and Sugiura, Komei}, volume = {100}, series = {Proceedings of Machine Learning Research}, month = {30 Oct--01 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v100/capotondi20a/capotondi20a.pdf}, url = {https://proceedings.mlr.press/v100/capotondi20a.html}, abstract = {By exploiting an a-priori estimate of the dynamic model of a manipulator, it is possible to command joint torques which ideally realize a Feedback Linearization (FL) controller. The exact cancellation may nevertheless not be achieved due to model uncertainties and possible errors in the estimation of the dynamic coefficients. In this work, an online learning scheme for control based on FL is presented. By reading joint positions and joint velocities information only (without the use of any torque measurement), we are able to learn those model uncertainties and thus achieve perfect FL control. Simulations results on the popular KUKA LWR iiwa robot are reported to show the quality of the proposed approach.} }
Endnote
%0 Conference Paper %T An Online Learning Procedure for Feedback Linearization Control without Torque Measurements %A M. Capotondi %A G. Turrisi %A C. Gaz %A V. Modugno %A G. Oriolo %A A. De Luca %B Proceedings of the Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2020 %E Leslie Pack Kaelbling %E Danica Kragic %E Komei Sugiura %F pmlr-v100-capotondi20a %I PMLR %P 1359--1368 %U https://proceedings.mlr.press/v100/capotondi20a.html %V 100 %X By exploiting an a-priori estimate of the dynamic model of a manipulator, it is possible to command joint torques which ideally realize a Feedback Linearization (FL) controller. The exact cancellation may nevertheless not be achieved due to model uncertainties and possible errors in the estimation of the dynamic coefficients. In this work, an online learning scheme for control based on FL is presented. By reading joint positions and joint velocities information only (without the use of any torque measurement), we are able to learn those model uncertainties and thus achieve perfect FL control. Simulations results on the popular KUKA LWR iiwa robot are reported to show the quality of the proposed approach.
APA
Capotondi, M., Turrisi, G., Gaz, C., Modugno, V., Oriolo, G. & Luca, A.D.. (2020). An Online Learning Procedure for Feedback Linearization Control without Torque Measurements. Proceedings of the Conference on Robot Learning, in Proceedings of Machine Learning Research 100:1359-1368 Available from https://proceedings.mlr.press/v100/capotondi20a.html.

Related Material