Sequential Monte Carlo in reachability heuristics for probabilistic planning

Bryce, Daniel; Kambhampati, Subbarao; Smith, David E.

doi:10.1016/j.artint.2007.10.018

Cited by 10 publications

(14 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…CPP is well studied in the AI literature (see, e.g., [2][3][4][5][6]). The probability of achieving a goal state t by executing a plan π = a 1 • • • a n is calculated by the following way (cf.…”

Section: Background Of Automated Planningmentioning

confidence: 99%

“…Although we have said that there are limitations of classical planning, there is an important line of work trying to transform conformant planning problems and even conformant probabilistic planning problems into classical planning problems w.r.t. different subsets of the original initial uncertainty set with high enough probability (see, e.g., [23,6]). The fully faithful transformations usually requires high computational costs, but we can find some sound but incomplete transformations which can work well in practice.…”

Section: Compilation To Classical Planningmentioning

confidence: 99%

See 1 more Smart Citation

A dynamic epistemic framework for reasoning about conformant probabilistic plans

Kooi

Wang

2019

Artificial Intelligence

View full text Add to dashboard Cite

In this paper, we introduce a probabilistic dynamic epistemic logical framework that can be applied for reasoning and verifying conformant probabilistic plans in a single agent setting. In conformant probabilistic planning (CPP), we are looking for a linear plan such that the probability of achieving the goal after executing the plan is no less than a given threshold probability δ. Our logical framework can trace the change of the belief state of the agent during the execution of the plan and verify the conformant plans. Moreover, with this logic, we can enrich the CPP framework by formulating the goal as a formula in our language with action modalities and probabilistic beliefs. As for the main technical results, we provide a complete axiomatization of the logic and show the decidability of its validity problem.

show abstract

Section: Background Of Automated Planningmentioning

confidence: 99%

Section: Compilation To Classical Planningmentioning

confidence: 99%

A dynamic epistemic framework for reasoning about conformant probabilistic plans

Kooi

Wang

2019

Artificial Intelligence

View full text Add to dashboard Cite

show abstract

“…The most effective heuristics for CPP involve estimating the cost to achieve the goal with probability no less than τ [15,16]. We employ the McLUG technique described by Bryce et al [15] to compute relaxed plans, using the number of actions as the heuristic.…”

Section: Heuristics For Cppmentioning

confidence: 99%

“…We employ the McLUG technique described by Bryce et al [15] to compute relaxed plans, using the number of actions as the heuristic. The approach taken by Bryce et al [15] to compute the heuristic for a belief state and value of τ is to compute k deterministic planning graphs and extract a relaxed plan that achieves the goals in at least kτ of the planning graphs. Each planning graph is deterministic because it is built with respect to a state sampled from the belief state and sampled outcomes of each probabilistic action in each action layer.…”

Section: Heuristics For Cppmentioning

confidence: 99%

See 1 more Smart Citation

Planning for Multiple Preferences versus Planning with No Preference

Bryce

2012

ISRN Artificial Intelligence

Self Cite

View full text Add to dashboard Cite

Many planning applications must address conflicting plan objectives, such as cost, duration, and resource consumption, and decision makers want to know the possible tradeoffs. Traditionally, such problems are solved by invoking a single-objective algorithm (such as A * ) on multiple, alternative preferences of the objectives to identify nondominated plans. The less-popular alternative is to delay such reasoning and directly optimize multiple plan objectives with a search algorithm like multiobjective A * (MOA * ). The relative performance of these two approaches hinges upon the number of f -values computed for individual search nodes. A * may revisit a node several times and compute a different f -value each time. MOA * visits each node once and may compute some number of f -values (each estimating the value of a different nondominated solution constructed from the node). While A * does not share f -values between searches for different solutions, MOA * can sometimes find multiple solutions while computing a single f -value per node. The results of extensive empirical comparison show that (i) the performance of multiple invocations of a single-objective A * versus a single invocation of MOA * is often worse in time and quality and (ii) that techniques for balancing per node cost and exploration are promising.

show abstract