Composable Energy Policies for Reactive Motion Generation and Reinforcement Learning.
Julen UrainPuze LiuAnqi LiCarlo D'EramoJan PetersPublished in: Robotics: Science and Systems (2021)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- image sequences
- motion analysis
- motion estimation
- markov decision process
- machine learning
- motion detection
- control policies
- energy consumption
- motion model
- space time
- spatial and temporal
- energy minimization
- motion capture
- function approximation
- markov decision problems
- motion tracking
- reward function
- reinforcement learning agents
- learning algorithm
- motion segmentation
- optical flow
- state space
- markov decision processes
- motion field
- human motion
- partially observable
- multi agent
- motion parameters
- reinforcement learning algorithms
- motion patterns
- temporal difference
- camera motion
- action selection
- agent architecture
- partially observable markov decision processes
- model free
- motion control
- dynamic programming
- long run
- continuous state
- moving objects
- motion planning