
TL;DR
This paper presents a goal-directed planning approach that incorporates external events with probabilistic occurrence, using Bayesian networks, Monte Carlo simulation, and Markov chains to evaluate plan success under uncertainty.
Contribution
It introduces a novel planning methodology that models external events with probabilities and employs Bayesian belief nets, Monte Carlo simulation, and Markov chains for success estimation.
Findings
Bayesian belief nets effectively model plan success probabilities.
Monte Carlo simulation provides a practical approximation method.
Markov chain approach leverages domain dependencies for efficiency.
Abstract
I describe a planning methodology for domains with uncertainty in the form of external events that are not completely predictable. The events are represented by enabling conditions and probabilities of occurrence. The planner is goal-directed and backward chaining, but the subgoals are suggested by analyzing the probability of success of the partial plan rather than being simply the open conditions of the operators in the plan. The partial plan is represented as a Bayesian belief net to compute its probability of success. Since calculating the probability of success of a plan can be very expensive I introduce two other techniques for computing it, one that uses Monte Carlo simulation to estimate it and one based on a Markov chain representation that uses knowledge about the dependencies between the predicates describing the domain.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAI-based Problem Solving and Planning · Logic, Reasoning, and Knowledge · Bayesian Modeling and Causal Inference
