Abstract
Mobile robots are enjoying increasing popularity in a number of different automation tasks. Omnidirectional mobile robots especially allow for a very flexible operation. They are able to accelerate in every direction, regardless of their orientation. In this context, we developed our own robot platform for research on said types of robots. It turns out that these mobile robots show interesting behaviour, which commonly used models for omnidirectional mobile robots fail to reproduce. As the exact sources and structures of mismatches are still unknown, non-parametric Gaussian process regression is used to develop a data-based model extension of the robot. A common control task for industrial applications is trajectory tracking, where a robot needs to follow a predefined path, for example in a warehouse, as close as possible in space and time. Appropriate feed-forward solutions for the data-based model are developed and finally leveraged in closed-loop control via nonlinear model predictive control. In real-world experiments, the results are compared to commonly used proportional position-based feedback. This novel contribution builds upon the preliminary work in [7] but, for the first time, includes also closed-loop (trajectory) tracking.
Zusammenfassung
Mobile Roboter erfreuen sich steigender Beliebtheit in einer Vielzahl von industriellen Anwendungen. Speziell omnidirektionale Roboter da diese, unabhängig von ihrer aktuellen Orientierung im Raum, in der Lage sind in jede Raumrichtung zu beschleunigen und daher flexibel einsetzbar sind. Vor diesem Hintergrund wurde eine eigene Roboterplattform entwickelt. Diese Roboter zeigen ein interessantes Verhalten, welches klassische Modellierungsansätze aus der Robotik nicht abbilden können. Da die genauen Fehlerquellen und seine Struktur nicht genauer bekannt sind, wird ein datenbasierter Korrekturterm auf Grundlage von Gauß-Prozessen entwickelt. Ein häufiger Anwendungsfall in industriellen Anwendungen ist die Trajektorienfolgeregelung. Hier muss der Roboter, beispielsweise in einer Lagerhalle, einer vordefinierten Trajektorie in Raum und Zeit folgen. Es werden geeignete Vorsteuerungen auf Grundlage des datenbasierten Modells entworfen und schließlich für die nichtlineare modellprädiktive Regelung verwendet. In Experimenten mit dem echten Roboter werden die Ergebnisse validiert und mit denen eines häufig verwendeten proportionalen Reglers auf Positionsebene verglichen. Dieser neuartige Beitrag fußt auf der vorläufigen Arbeit aus [7], beinhaltet aber erstmals auch die Trajektorienfolgeregelung.
Funding source: Deutsche Forschungsgemeinschaft
Award Identifier / Grant number: EXC 2075 – 390740016
Award Identifier / Grant number: project PN4-4
Award Identifier / Grant number: EB195/32-1
Award Identifier / Grant number: 433183605
Funding statement: This research is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy – EXC 2075 – 390740016, project PN4-4 “Theoretical Guarantees for Predictive Control in Adaptive Multi-Agent Scenarios” and project EB195/32-1, 433183605 “Research on Multibody Dynamics and Control for Collaborative Elastic Object Transportation by a Heterogeneous Swarm with Aerial and Land-Based Mobile Robots”.
About the authors
Hannes Eschmann received the B. Sc. degree in mechanical engineering and the M. Sc. degree in theoretical mechanical engineering from the Hamburg University of Technology, Germany, in 2016 and 2019, respectively. He is currently pursuing the doctoral degree with the Institute of Engineering and Computational Mechanics (ITM), University of Stuttgart, Germany. His research interests include the application of control theory and machine learning in mobile robotics, especially robust and distributed model predictive control and Gaussian process regression.
Henrik Ebel received his B. Sc. and M. Sc. degrees in simulation technology from the University of Stuttgart, Germany, in 2014 and 2016. He is currently a doctoral student and a member of the research staff at the Institute of Engineering and Computational Mechanics at the University of Stuttgart. His research interests include control theory, multibody system dynamics, and robotics. Of particular interest are the cooperation of multiple robotic agents, as well as optimization-based control schemes.
Peter Eberhard is full professor and since 2002 director of the Institute of Engineering and Computational Mechanics (ITM) at the University of Stuttgart, Germany. He was Treasurer and Bureau member of IUTAM, the International Union of Theoretical and Applied Mechanics, and served before in many national and international organizations, e. g., as Chairman of the IMSD (International Association for Multibody System Dynamics) or DEKOMECH (German Committee for Mechanics).
References
1. J. A. E. Andersson, J. Gillis, G. Horn, J. B. Rawlings and M. Diehl. CasADi: a software framework for nonlinear optimization and optimal control. Mathematical Programming Computation, 11(1):1–36, 2019.10.1007/s12532-018-0139-4Search in Google Scholar
2. M. Bauer, M. van der Wilk and C. E. Rasmussen. Understanding probabilistic sparse Gaussian process approximations. In Advances in Neural Information Processing Systems, Barcelona, Spain, volume 29, pages 1533–1541, 2016.Search in Google Scholar
3. S. L. Brunton, J. L. Proctor and J. N. Kutz. Sparse identification of nonlinear dynamics with control (SINDYc). IFAC-PapersOnLine, 49(18):710–715, 2016.10.1016/j.ifacol.2016.10.249Search in Google Scholar
4. H. Ebel and P. Eberhard. Optimization-driven control and organization of a robot swarm for cooperative transportation. IFAC-PapersOnLine, 52(15):115–120, 2019.10.1016/j.ifacol.2019.11.660Search in Google Scholar
5. H. Ebel and P. Eberhard. A comparative look at two formation control approaches based on optimization and algebraic graph theory. Robotics and Autonomous Systems, 136:103686, 2021.10.1016/j.robot.2020.103686Search in Google Scholar
6. H. Ebel, W. Luo, F. Yu, Q. Tang and P. Eberhard. Design and experimental validation of a distributed cooperative transportation scheme. IEEE Transactions on Automation Science and Engineering, pages 1–13, 2020. DOI: 10.1109/TASE.2020.2997411.Search in Google Scholar
7. H. Eschmann, H. Ebel and P. Eberhard. Data-based model of an omnidirectional mobile robot using Gaussian processes. In IFAC Symposium on System Identification (SYSID) – Learning models for decision and control, Padova, Italy, pages 1–6, 2021. Accepted for publication.10.1016/j.ifacol.2021.08.327Search in Google Scholar
8. H. Eschmann and P. Eberhard. Learning‐based model predictive control for multi‐agent systems using Gaussian processes. In Proceedings in Applied Mathematics and Mechanics (PAMM), Kassel, Germany, volume 1, pages 1–2, 2021.10.1002/pamm.202000009Search in Google Scholar
9. H. Ferreau, C. Kirches, A. Potschka, H. Bock and M. Diehl. qpOASES: A parametric active-set algorithm for quadratic programming. Mathematical Programming Computation, 6(4):327–363, 2014.10.1007/s12532-014-0071-1Search in Google Scholar
10. L. Hewing, J. Kabzan and M. N. Zeilinger. Cautious model predictive control using Gaussian process regression. IEEE Transactions on Control Systems Technology, 28(6):2736–2743, 2020.10.1109/TCST.2019.2949757Search in Google Scholar
11. M. D. McKay, R. J. Beckman and W. J. Conover. A comparison of three methods for selecting values of input variables in the analysis of output from a computer code. Technometrics, 21(2):239–245, 1979.10.1080/00401706.1979.10489755Search in Google Scholar
12. C. J. Ostafew, A. P. Schoellig and T. D. Barfoot. Robust constrained learning-based NMPC enabling reliable mobile robot path tracking. The International Journal of Robotics Research, 35(13):1547–1563, 2016.10.1177/0278364916645661Search in Google Scholar
13. J. Quiñonero Candela and C. E. Rasmussen. A unifying view of sparse approximate Gaussian process regression. Journal of Machine Learning Research, 6:1939–1959, 2005.Search in Google Scholar
14. C. E. Rasmussen and H. Nickisch. Gaussian processes for machine learning (GPML) toolbox. Journal of Machine Learning Research, 11:3011–3015, 2010.Search in Google Scholar
15. C. E. Rasmussen and C. K. I. Williams. Gaussian Processes for Machine Learning (Adaptive Computation and Machine Learning). The MIT Press, Cambridge, Massachusetts, 2005.10.7551/mitpress/3206.001.0001Search in Google Scholar
16. J. Rawlings, D. Q. Mayne and M. Diehl. Model Predictive Control: Theory, Computation, and Design. Nob Hill Publishing, 2nd edition, 2017.Search in Google Scholar
17. E. Snelson and Z. Ghahramani. Sparse Gaussian processes using pseudo-inputs. In Proceedings of the 18th International Conference on Neural Information Processing Systems, Vancouver, Canada, pages 1257–1264, 2005.Search in Google Scholar
18. M. Titsias. Variational learning of inducing variables in sparse Gaussian processes. In Proceedings of Machine Learning Research, Clearwater Beach, USA, volume 5, pages 567–574, 2009.Search in Google Scholar
19. A. Wächter and L. T. Biegler. On the implementation of an interior-point filter line-search algorithm for large-scale nonlinear programming. Mathematical Programming, 106(1):25–57, 2006.10.1007/s10107-004-0559-ySearch in Google Scholar
20. I. Zeidis and K. Zimmermann. Dynamics of a four-wheeled mobile robot with Mecanum wheels. Journal of Applied Mathematics and Mechanics, 99(12):1–22, 2019.10.1002/zamm.201900173Search in Google Scholar
© 2021 Walter de Gruyter GmbH, Berlin/Boston
