Efficient optimisation algorithms for real-time Model Predictive Control
Project Leader: Ye Pu
Primary Contact: Ye Pu (email@example.com)
Keywords: autonomous systems; real-time control; real-time optimisation
Disciplines: Electrical & Electronic Engineering
Real-time optimal control and estimation tackles the problems of computing stabilising controllers or estimators with limited resources, eg, computation time, data storage space or communication channels.
In this project, we focus on developing fast optimisation algorithms to solve real-time control and estimation problems. Specifically, we aim to design algorithms which do not attempt to give an optimal solution with high accuracy, but only provide feasible/sub-optimal solutions which are good enough for specific applications. In other words, I consider the online resources such as computation time as a part of the specification for controller design. The algorithms aim at finding an ‘OK’ solution with only a small amount of resources, while at the same time give performance guarantee for the closed-loop systems. In particular, we will study the tradeoff between different system specifications with given resources, as well as the optimal resource allocation problem.