A belief-desire-intention architechture with a logic-based planner for agents in stochastic domains

This dissertation investigates high-level decision making for agents that are both goal and utility \ndriven. We develop a partially observable Markov decision process (POMDP) planner which \nis an extension of an agent programming language called DTGolog, itself an extension of the \nGolog language. Golog is based on a logic for reasoning about action—the situation calculus. \nA POMDP planner on its own cannot cope well with dynamically changing environments \nand complicated goals. This is exactly a strength of the belief-desire-intention (BDI) model: \nBDI theory has been developed to design agents that can select goals intelligently, dynamically \nabandon and adopt new goals, and yet commit to intentions for achieving goals. The contribution \nof this research is twofold: (1) developing a relational POMDP planner for cognitive \nrobotics, (2) specifying a preliminary BDI architecture that can deal with stochasticity in action \nand perception, by employing the planner.