This unit introduces engineering design via optimisation, i. e. finding the "best possible" solution to a particular problem. For example, an autonomous vehicle must find the fastest route between two locations over a road network; a biomedical sensing device must compute the most accurate estimate of important physiological parameters from noise-corrupted measurements; a feedback control system must stabilise and control a multivariable dynamical system (such as an aircraft) in an optimal fashion. The student will learn how to formulate a design in terms of a "cost function", when it is possible to find the "best" design via minimization of this "cost", and how to do so. The course will introduce widely-used optimisation frameworks including linear and quadratic programming (LP and QP), dynamic programming (DP), path planning with Dijkstra's algorithm, A*, and probabilistic roadmaps (PRMs), state estimation via Kalman filters, and control via the linear quadratic regulator (LQR) and Model Predictive Control (MPC). There will be constant emphasis on connections to real-world engineering problems in control, robotics, aerospace, biomedical engineering, and manufacturing.
Lectures, Tutorials, Research
Through semester assessment (50%) and Final Exam (50%)
Strong understanding of feedback control systems, specifically in the area of system modelling and control design in the frequency domain.