Planning in Factored Action Spaces with Symbolic Dynamic Programming.
Aswin RaghavanSaket JoshiAlan FernPrasad TadepalliRoni KhardonPublished in: AAAI (2012)
Keyphrases
- action space
- state space
- dynamic programming
- planning problems
- heuristic search
- markov decision problems
- markov decision processes
- state and action spaces
- continuous state
- reinforcement learning
- partially observable
- optimal policy
- skill learning
- action selection
- control policies
- state variables
- orders of magnitude
- reinforcement learning problems
- markov chain
- linear programming
- initial state
- partially observable markov decision processes
- real valued
- finite state
- optimal control
- belief state
- continuous action
- particle filter
- partially observable markov decision process
- dynamical systems
- stereo matching
- planning under uncertainty
- markov decision process
- stochastic processes
- planning domains
- motion planning
- dynamic environments
- robot arm
- image segmentation
- reinforcement learning algorithms
- mobile robot
- cooperative
- infinite horizon