Motion Perception in Reinforcement Learning with Dynamic Objects.
Artemij AmiranashviliAlexey DosovitskiyVladlen KoltunThomas BroxPublished in: CoRR (2019)
Keyphrases
- reinforcement learning
- visual perception
- motion estimation
- space time
- motion analysis
- motion model
- image sequences
- function approximation
- visual motion
- optical flow
- motion parameters
- perceptual information
- motion patterns
- state space
- spatial and temporal
- model free
- human perception
- active exploration
- three dimensional
- learning process
- camera motion
- motion field
- machine learning
- motion tracking
- action selection
- transfer learning
- motion segmentation
- motion planning
- reference frame
- particle filter
- humanoid robot
- optimal control
- temporal difference
- point correspondences
- object motion
- image motion
- spatio temporal
- moving objects