Gradient-Based Mixed Planning with Discrete and Continuous Actions.
Kebing JinHankz Hankui ZhuoZhanhao XiaoHai WanSubbarao KambhampatiPublished in: CoRR (2021)
Keyphrases
- continuous action
- continuous domains
- stochastic domains
- discrete space
- partially observable markov decision processes
- action selection
- continuous variables
- decision theoretic
- policy search
- plan recognition
- continuous state
- continuous space
- action space
- discrete variables
- classical planning
- planning problems
- discrete data
- plan execution
- macro actions
- domain independent
- continuous functions
- goal directed
- concurrent actions
- continuous valued
- external events
- world model
- heuristic search
- continuous state spaces
- reinforcement learning
- partial observability
- sensing actions
- finite number
- ai planning
- derived predicates
- discrete valued
- continuous data
- incomplete knowledge
- plan generation
- initial state
- partially observable
- motion planning
- human actions
- decision support
- video sequences