ROLL: Visual Self-Supervised Reinforcement Learning with Object Reasoning.
Yufei WangGautham Narayan NarasimhanXingyu LinBrian OkornDavid HeldPublished in: CoRL (2020)
Keyphrases
- reinforcement learning
- visual objects
- d objects
- visual properties
- knowledge representation
- spatial relations
- multiple objects
- visual appearance
- function approximation
- optimal control
- high speed
- knowledge base
- object tracking
- reasoning systems
- learning process
- reasoning process
- complex objects
- object level
- real world objects
- data objects
- spatial configurations
- spatial relationships
- machine learning
- markov decision processes
- visual information
- optimal policy
- moving objects
- computer vision
- learning algorithm