The CL-Buridan System: Probabilistic Planning with Incomplete Information

This paper describes (,:LBURl DAN an implementf'd planner for problem domains in whicb tbe agent is uncertain about the initial world state and the effects of its own actions, but has sensors that allow it to improve its state of information. The system uses a probabilistir semantics to represent incomplete information, and provid('s for actions with informational as well as causal effects. The action representation allows an action's causal and information al effpcts to be freely mixed, and can I'f'pTPsent sensors whose informational content is noisy and state dependent. Information obtained at run time can be exploitpd by conditional and looping constructs in the plan language. This paper describ es the basic structure of the CLB l' RI DA 'l representation and algorithm, and also provide an analysis of assumptions about tar!2;et problem domains that would makP thpm an (i.ppropriate (or inappropriate) choice fur using this technology to build a problf'm-solving agent.