My research interests
During my PhD I attempted to revive partial-order planning (POP), a branch of planning technology that was superseded by forward-chaining planning. Sadly, my attempts to improve the heuristic search and utilising Landmarks did not revive POP. However, this research did infused my interest in lifted planning and ultimately lead to the lifted planner L-RPG. This planner can solve much larger problem instances than any state-of-the-art planner.
When researching Contingency Planning I found that most planners make an implicit assumption that all observations are correlated. I found a method to factorise observations, leading to much better performance and better scaling.
SQUIRREL is the latest project I am involved with and I am now working on how to use planning technology for social interaction with children.