Physically Embedded Planning Problems: New Challenges for Reinforcement Learning.
Mehdi MirzaAndrew JaegleJonathan J. HuntArthur GuezSaran TunyasuvunakoolAlistair MuldalThéophane WeberPéter KarkusSébastien RacanièreLars BuesingTimothy P. LillicrapNicolas HeessPublished in: CoRR (2020)
Keyphrases
- planning problems
- reinforcement learning
- state space
- deterministic domains
- partial observability
- continuous state
- domain independent
- heuristic search
- fully observable
- ai planning
- planning systems
- orders of magnitude
- optimal planning
- planning domains
- solving planning problems
- probabilistic planning
- learning algorithm
- machine learning
- classical planning
- domain independent planning
- action space
- concurrent actions
- stochastic domains
- htn planning
- hidden state
- function approximation
- reinforcement learning algorithms
- optimal policy
- model free
- search space
- conformant planning
- causal graph
- multi agent
- partially observable
- dynamic programming
- forward search
- binary decision diagrams
- markov decision process