Codesign of Optimisation-Based Controllers on FPGAs

The performance of a closed-loop system is a function not only of the controller, but also of the physical system parameters and computer architecture. Careful codesign of all subsystems can have a dramatic impact on the performance and robustness of the overall cyber-physical system, especially when implementing computationally demanding optimisation-based control methods. In this presentation, Eric outlines how Imperial College London has been integrating MATLAB^® and Simulink^® as part of their design process for hardware-in-the-loop testing, where a predictive controller is implemented in real-time on an FPGA target. They have used Model-Based Design to not only reduce costs and design time, but also achieve performance specifications that are one or more orders of magnitude beyond the state-of-the-art.