Emergence of flexible prediction-based discrete decision making and continuous motion generation through actor-Q-learning.
Katsunari ShibataKenta GotoPublished in: ICDL-EPIROB (2013)
Keyphrases
- decision making
- continuous state spaces
- continuous state and action spaces
- prediction accuracy
- action selection
- discrete space
- motion prediction
- prediction model
- space time
- motion estimation
- continuous functions
- continuous variables
- reinforcement learning
- continuous domains
- cooperative
- decision support system
- decision makers
- continuous space
- state space
- continuous action
- multi agent
- function approximation
- continuous data
- learning algorithm
- sequence prediction
- human motion
- motion analysis
- decision support
- camera motion
- image sequences
- discrete data
- grey prediction model
- motion field
- continuous valued
- optical flow
- fuzzy logic
- markov decision processes
- motion detection
- motion model
- reinforcement learning algorithms
- data mining
- discrete valued
- moving objects
- prediction error
- decision process
- supply chain
- dynamic environments
- stochastic approximation
- discrete variables
- humanoid robot
- spatio temporal
- video sequences
- least squares
- motion capture