Predictive control

Prof Colin Jones

We study the theory and practice of model predictive control applied to constrained systems, with an emphasis on problems arising from renewable power. We are particularly interested in the following topics:

Crystallization : Optimal Control and Advanced Monitoring

Swiss National Science Foundation funded project. 2011 – 2014

Collaborative work with:

EPFL project team: Jean-Hubert Hours and Colin Jones

CrystOCAM project website

Advanced Vision Algorithms and Optimal Monitoring for Crystallization Processes

Vision methods are a promising tool for describing and measuring the particle size distribution of crystals in solution. We investigate advanced vision algorithms capable of estimating the three-dimensional shape of individual crystals from stereo images under the microscope. Part of the work consists in developing global optimization methods for computing the 3D pose of a crystal through rotation space search. As a second step, we will investigate strategies to control the final particle size distribution of the crystal ensemble.


Embedded Optimization for Resource-Constrained Platforms

European FP7 Project. 2010 – 2013

EPFL project team: Melanie Zeilinger and Colin Jones

Project website

There is enormous economic potential for the application of embedded optimization technologies in embedded systems design. Recent advances in the performance of embedded hardware platforms, in combination with fundamental improvements in optimization theory and algorithms, have opened the door to widespread applications over the next decade.
Embedded optimization will enable huge energy and resource savings, increased safety, and improved fault detection across a wide a range of industrial applications in the mechatronic, automotive, process control and aerospace sectors. In order to realize the full potential of optimization in embedded systems, their design must also be supported by a focussed set of tools enabling the rapid transfer of novel high-performance algorithms to practical applications.

The EMBOCON consortium will enable widespread application of real-time optimization in embedded systems through:

  • Tailoring of customized numerical algorithms to increase their robustness and efficiency on embedded systems
  • Enabling real-time optimization on cheap industry-standard hardware platforms
  • Defining a common user interface for optimization technologies to facilitate technology transfer to industry
  • Performing challenging case studies in cooperation with industrial partners to demonstrate technological maturity.

The EMBOCON consortium will strengthen a network of world-leading academic and industrial partners with complementary expertise in control, optimization and embedded systems in a range of industrial applications. Particular emphasis is placed on close collaboration between mathematical algorithm developers, control theorists, hardware specialists and industrial application engineers. The network will consolidate and extend Europe’s position as the world research leader in these areas and foster strong collaborative links between European academia and industry.