Cost-Sensitive Reachability Heuristics for Probabilistic Planning
نویسنده
چکیده
Reachability heuristics have lead to impressive scale-ups in deterministic planning making their application to probabilistic planning a promising research direction. We describe how one such reachability heuristic (based on planning graphs) can be extended to handle a class of cost-sensitive probabilistic planning problems. Specifically, we address the problem of conformant (non-observable) probabilistic planning, which is a special case of the non-observable MDP (NOMDP) problem. We show how the quality of plans generated by our planner improves with our cost-sensitive heuristic and how our planner far out-scales the best existing probabilistic planner.
منابع مشابه
Probabilistic Planning is Multi-objective!
Probabilistic planning is an inherently multi-objective problem where plans must trade-off probability of goal satisfaction with expected plan cost. To date, probabilistic plan synthesis algorithms have focussed on single objective formulations that bound one of the objectives by making some unnatural assumptions. We show that a multi-objective formulation is not only needed, but also enables u...
متن کاملSCALABLE PLANNING UNDER UNCERTAINTY by
Autonomous agents that act in the real-world can often improve their success by capturing the uncertainty that arises because of their imperfect knowledge and potentially faulty actions. By making plans robust to uncertainty, agents can be prepared to counteract plan failure or act upon information that becomes available during plan execution. Such robust plans are valuable, but are often diffi...
متن کاملSequential Monte Carlo in Probabilistic Planning Reachability Heuristics
The current best conformant probabilistic planners encode the problem as a bounded length CSP or SAT problem. While these approaches can find optimal solutions for given plan lengths, they often do not scale for large problems or plan lengths. As has been shown in classical planning, heuristic search outperforms CSP/SAT techniques (especially when a plan length is not given a priori). The probl...
متن کاملNew Admissible Heuristics for Domain-Independent Planning
Admissible heuristics are critical for effective domainindependent planning when optimal solutions must be guaranteed. Two useful heuristics are the h heuristics, which generalize the reachability heuristic underlying the planning graph, and pattern database heuristics. These heuristics, however, have serious limitations: reachability heuristics capture only the cost of critical paths in a rela...
متن کاملSequential Monte Carlo in reachability heuristics for probabilistic planning
Some of the current best conformant probabilistic planners focus on finding a fixed length plan with maximal probability. While these approaches can find optimal solutions, they often do not scale for large problems or plan lengths. As has been shown in classical planning, heuristic search outperforms bounded length search (especially when an appropriate plan length is not given a priori). The ...
متن کامل