CN-SK cooperation: Robust Model Predictive Control Meets Robotics

Title: Robust Model Predictive Control Meets Robotics

Project code: APVV SK-CN-2015-0016

Partners:

Duration: 2016-2017

Abstract:

China and Slovakia have an enormous potential for innovative research in robotics and control. The goal of this project is to bring together a group of young researchers whose aim is to create robust model predictive controllers with emphasis on the implementation of advanced control procedures and applications in robotics. We will build upon advanced linear matrix inequality techniques and real-time control software to develop novel types of autonomous and intelligent control algorithms for uncertain processes that are far beyond the state-of-the-art. The research shall be carried out by scientist and PhD students in Slovakia and China, who will visit each other on a regular basis, thereby creating channels for technological as well as intercultural exchange. At the same time, we will showcase innovative research on modern technologies that shall educate the next generation of control and robotic scientists in China and Slovakia, thereby creating a huge potential for academic breakthroughs as well as successful spin-offs in both countries.

 

Project description:

Model predictive control (MPC) is an advanced control strategy that is widely used in industrial process control due to its ability to cope with pyhsical process models as well as models for the input- and output constraints. Recent advances in the field of embedded hardware platforms as well as fundamental improvements in the field of optimization algorithms have opened the door to widespread application to systems with fast sampling times as for example found in the mechatronic and automotive sectors. However, in order to realize the full potential of model predictive control, in particular, for embedded systems and autonomous robots, we first need reliable optimization and model predictive control algorithms that can deal with the often nonlinear, uncertainty affected, and highly dynamic nature of modern industrial processes and robots. Unfortunately, robust MPC methods, which can deal with uncertainties in the context of MPC, are not yet real-time feasible for systems with high sampling time. In order to remedy this situation, this project will bring together researches from ShanghaiTech and the Slovak University of Technology in Bratislava, who are experts in the field of fast and autogenerated MPC algorithms, linear matrix inequality (LMI) based robust MPC design, and application of MPC in robotics. The scientific goals of this collaboration are to:
1. synthesize advanced robust MPC design for uncertain dynamic systems
2. develop software packages that enable the transfer of theory based robust control design to high impact applications
3. perform challenging case studies by implementing robust MPC controllers for autonomous robots in the ShanghaiTech robotics and control lab in order to illustrate the maturity of the developments and to promote wide acceptance by industry.
The two groups will collaborate closely in order to jointly develop robust optimization and robust control design technologies, software, and robotic applications. All software will be made freely available and shall be tested on real-world robots. The results will be published in top-tier journals, at jointly organized conference sessions and workshops as well as on a tri-lingual project web page.

- Month 1-6: LMI based formulation of Robust MPC, organize joint conference session
- Month 7-12: joint work on a software implementation. For this aim the first batch of PhD students will be exchanged.
- Month 13-18: testing robust MPC on real-world robot hardware. For this aim the second batch of PhD students will be exchanged.
- Month 19-24: dissemination of research results at conferences, at least one mature journal publication.

The project aim to establish a collaborative research partnership between the Chinese and the Slovak teams using the existing common research interests in topics related to advanced robust MPC and robust optimization. The complementarity of the teams comes from the fact that the Slovak partners are focused on the optimal control design, MPC design, and robust MPC design, while the Chinese partners come with their experience in the robust optimization, nonlinear optimization and the well-equipped new robotic laboratories suitable for the experimental validation of the attained results. Moreover, the Chinese partners are the main developers of the ACADO Toolkit for automatic control and dynamic optimization. ACADO is written in C++ and there is also a MATLAB interface.

Publications

2017

  1. Ingole, D. – Kvasnica, M. – De Silva, H. – Gustafson, J.: Reducing Memory Footprints in Explicit Model Predictive Control using Universal Numbers. In Preprints of the 20th IFAC World Congress, Toulouse, France, vol. 20, pp. 12100–12105, 2017.
  2. Oravec, J. – Jiang, Y. – Houska, B. – Kvasnica, M.: Parallel Explicit MPC for Hardware with Limited Memory. In Preprints of the 20th IFAC World Congress, Toulouse, France, vol. 20, pp. 3356–3361, 2017.

2016

  1. Oravec, J. – Bakošová, M.: Soft Constraints in the Robust MPC Design via LMIs. In American Control Conference, Boston, Massachusetts, USA, pp. 3588–3593, 2016.
  2. Oravec, J. – Bakošová, M. – Mészáros, A. – Míková, N.: Experimental Investigation of Alternative Robust Model Predictive Control of a Heat Exchanger. Applied Thermal Engineering, pp. 774–782, 2016.
  3. Oravec, J. – Kalúz, M. – Bakaráč, P. – Bakošová, M.: Improvements of Educational Process of Automation and Optimization Using 2D Plotter. In Preprints of the 11th IFAC Symposium on Advances in Control Education, vol. 11, pp. 16–21, 2016.
  4. Picard, D. – Drgoňa, J. – Helsen, L. – Kvasnica, M.: Impact of the controller model complexity on MPC performance evaluation for building climate control. In The European Conference on Computational Optimization, Leuven, Belgium, vol. 4, 2016.
  5. Števek, J. – Fikar, M.: Teaching aids for laboratory experiments with AR.Drone2 quadrotor - extended version. 2016.
  6. Števek, J. – Fikar, M.: Teaching aids for laboratory experiments with AR.Drone2 quadrotor. In Preprints of the 11th IFAC Symposium on Advances in Control Education, vol. 11, pp. 236–241, 2016.
  7. Števek, J. – Katuščák, S. – Fikar, M. – Dubinyová, L.: An automatic identification of wood materials from color images. Editor(s): J. Cigánek (SK), Š. Kozák (SK), A. Kozáková (SK),, In 2016 Cybernetics & Informatics (K&I), 28th International Conference, Levoča, vol. 28, 2016.
Facebook / Youtube

Facebook / Youtube