A Practical Approach to Insertion with Variable Socket Position Using Deep Reinforcement Learning.
Mel VeceríkOleg SushkovDavid BarkerThomas RothörlTodd HesterJonathan ScholzPublished in: ICRA (2019)
Keyphrases
- reinforcement learning
- state space
- function approximation
- robotic control
- model free
- reinforcement learning algorithms
- learning process
- multi agent
- artificial neural networks
- multi agent reinforcement learning
- temporal difference
- position and orientation
- deep learning
- transition model
- action selection
- belief nets
- optimal policy
- markov decision processes
- machine learning
- control system
- search algorithm
- case study
- search engine
- learning algorithm
- genetic algorithm