How to pick the domain randomization parameters for sim-to-real transfer of reinforcement learning policies?
Quan VuongSharad VikramHao SuSicun GaoHenrik I. ChristensenPublished in: CoRR (2019)
Keyphrases
- reinforcement learning
- transfer learning
- optimal policy
- cross domain
- policy search
- privacy preserving
- learning algorithm
- state space
- maximum likelihood
- analytical models
- parameter values
- learning tasks
- partially observable domains
- deterministic domains
- temporal difference
- model free
- domain independent
- function approximation
- machine learning
- complex domains
- parameter estimation
- control policies
- domain specific
- hierarchical reinforcement learning
- dynamic programming
- transferring knowledge
- genetic algorithm