Progressive extension of reinforcement learning action dimension for asymmetric assembly tasks.
Yuhang GaiJiuming GuoDan WuKen ChenPublished in: CoRR (2021)
Keyphrases
- reinforcement learning
- action selection
- partially observable domains
- function approximation
- action space
- state space
- learning algorithm
- reward shaping
- model free
- fitted q iteration
- temporal difference learning
- initial state
- reinforcement learning algorithms
- human actions
- optimal policy
- learning capabilities
- sensory inputs
- multiple dimensions
- agent learns
- markov decision processes
- agent receives
- supervised learning