Michael's research explores how interactive visualisation can be used to understand complex data and build user-guided AI systems. His research explores automatic and user-guided approaches for the layout of diagrams and documents. These approaches aim to combine the power and flexibility of optimisation or constraint-solving techniques with insights and guidance provided by expert domain users. This research is used to produce better layout tools in interactive diagram editors, for high-quality routing in commercial circuit diagramming software, for visualisation of automated layout and pipe routing for Liquefied Natural Gas (LNG) plants, and for automatic layout of protein structure diagrams and other biological networks.