Bridging Commonsense Reasoning and Probabilistic Planning via a Probabilistic Action Language

  • 2019-07-31 15:29:44
  • Yi Wang, Shiqi Zhang, Joohyung Lee
  • 2

Abstract

To be responsive to dynamically changing real-world environments, anintelligent agent needs to perform complex sequential decision-making tasksthat are often guided by commonsense knowledge. The previous work on this lineof research led to the framework called "interleaved commonsense reasoning andprobabilistic planning" (icorpp), which used P-log for representingcommmonsense knowledge and Markov Decision Processes (MDPs) or PartiallyObservable MDPs (POMDPs) for planning under uncertainty. A main limitation oficorpp is that its implementation requires non-trivial engineering efforts tobridge the commonsense reasoning and probabilistic planning formalisms. In thispaper, we present a unified framework to integrate icorpp's reasoning andplanning components. In particular, we extend probabilistic action languagepBC+ to express utility, belief states, and observation as in POMDP models.Inheriting the advantages of action languages, the new action language providesan elaboration tolerant representation of POMDP that reflects commonsenseknowledge. The idea led to the design of the system pbcplus2pomdp, whichcompiles a pBC+ action description into a POMDP model that can be directlyprocessed by off-the-shelf POMDP solvers to compute an optimal policy of thepBC+ action description. Our experiments show that it retains the advantages oficorpp while avoiding the manual efforts in bridging the commonsense reasonerand the probabilistic planner.

 

Quick Read (beta)

This feature is not avaialbe for this paper.