1 / 9

Two Models of Evaluating Probabilistic Planning

This document explores two models of evaluating probabilistic planning: the IPPC and FF strategies. It discusses how frequently goals are achieved under specified time constraints, focusing on the efficacy of FF-HOP and FF-Replan approaches. Key evaluation metrics include the quality of the policy and the speed of convergence to an optimal policy, highlighted through methods such as LRTDP and mGPT. The text delves into Kolobov’s heuristics for stochastic planning, providing insights into relaxation techniques for interactions and uncertainty, as well as the implications of determinization on planning efficacy.

onslow
Télécharger la présentation

Two Models of Evaluating Probabilistic Planning

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Two Models of Evaluating Probabilistic Planning • IPPC (Probabilistic Planning Competition) • How often did you reach the goal under the given time constraints • FF-HOP • FF-Replan • Evaluate on the quality of the policy • Converging to optimal policy faster • LRTDP • mGPT • Kolobov’s approach

  2. Heuristics for Stochastic Planning • Heuristics come from relaxation • We can relax along two separate dimensions: • Relax –ve interactions • Consider +ve interactions alone using relaxed planning graphs • Relax uncertainty • Consider determinizations • Or a combination of both!

  3. Determinizations • Most-likely outcome determinization • Inadmissible • e.g. if only path to goal relies on less likely outcome of an action • All outcomes determinization • Admissible, but not very informed • e.g. Very unlikely action leads you straight to goal

  4. Solving Determinizations • If we relax –ve interactions • Then compute relaxed plan • Admissible if optimal relaxed plan is computed • Inadmissible otherwise • If we keep –ve interactions • Then use a deterministic planner (e.g. FF/LPG) • Inadmissible unless the underlying planner is optimal

  5. Dimensions of Relaxation 3 4 Negative Interactions Increasing consideration  1 2 Uncertainty Relaxed Plan Heuristic 1 Reducing Uncertainty Bound the number of stochastic outcomes  Stochastic “width” McLUG 2 FF/LPG 3 4 Limited width stochastic planning?

  6. Dimensions of Relaxation Uncertainty -ve interactions

  7. Expressiveness v. Cost Node Expansions v. Heuristic Computation Cost Limited width stochastic planning FF McLUG Nodes Expanded FF-Replan Computation Cost h = 0 FFR FF

  8. Reducing Heuristic Computation Cost • Exploit overlapping structure of heuristics for different states • E.g. SAG idea for McLUG • E.g. Triangle tables idea for plans (c.f. Kolobov)

  9. Triangle Table Memoization • Use triangle tables / memoization C B A A B C If the above problem is solved, then we don’t need to call FF again for the below: B A A B

More Related