In the area of AI planning, I have in particular focused on plan recognition. Plan recognition involves inferring the intentions of an agent from a set of observations. Generally speaking, systems that attempt to interact with people in an intelligent and co-operative manner need to know many things about the individuals with whom they are interacting. Especially, it is a common problem for many intelligent co-operative systems to determine what the user really wants to achieve by using the system because he/she frequently does not state his/her goals explicitly. Sometimes the system also needs to know how the user plans to accomplish his/her goal.
I have recently developed a novel approach based on a two–stage paradigm of graph construction and analysis. This approach explicitly constructs a graph structure, called a Goal Graph, which is then analysed to recognise goals and plans. My attempt to use graph construction and analysis for plan recognition was in spirit influenced by Blum and Furst’s work on planning with Planning Graphs, at Carnegie Mellon University in late nineties. Blum and Furst’s work has been widely regarded as a major break-through in AI planning. My approach to plan recognition can be seen as a counterpart of Blum and Furst’s approach to planning. My research has led to a number of publications in the most prestigious international journals and conferences on AI, e.g., JAIR, AAAI, ECAI, and a book chapter in Current Trends in AI Planning. The research has also led to three research grants.