Learning to Touch Objects Through Stage-Wise Deep Reinforcement Learning.
François De La BourdonnayeCéline TeulièreJochen TrieschThierry ChateauPublished in: IROS (2018)
Keyphrases
- reinforcement learning
- learning problems
- learning algorithm
- learning process
- function approximation
- learning stage
- knowledge acquisition
- pairwise
- neural network
- autonomous learning
- supervised learning
- online learning
- robot control
- eligibility traces
- evolutionary learning
- reinforcement learning methods
- learning agent
- previously unseen
- learning tasks
- d objects
- dynamic programming
- prior knowledge