Offline Goal-Conditioned Reinforcement Learning for Shape Control of Deformable Linear Objects.
Rita LaezzaMohammadreza Shetab-BushehriGabriel Arslan WalterssonErol ÖzgürYoucef MezouarYiannis KarayiannidisPublished in: CoRR (2024)
Keyphrases
- reinforcement learning
- partially occluded objects
- control problems
- optimal control
- articulated objects
- real time
- object shape
- shape model
- object shapes
- complex shapes
- deformable shapes
- function approximation
- visual appearance
- arbitrarily shaped
- robot control
- control system
- relative position
- shape variability
- focus of attention
- machine learning
- topological properties
- state space
- d objects
- shape descriptors
- shape analysis
- medical images
- reinforcement learning algorithms
- learning algorithm
- object recognition
- invariant properties
- object model
- shape representation
- deformable models
- object features
- control policy
- control strategy
- action selection
- temporal difference