PlayVirtual: Augmenting Cycle-Consistent Virtual Trajectories for Reinforcement Learning.
Tao YuCuiling LanWenjun ZengMingxiao FengZhizheng ZhangZhibo ChenPublished in: NeurIPS (2021)
Keyphrases
- reinforcement learning
- augmented reality
- function approximation
- virtual environment
- reinforcement learning algorithms
- markov decision processes
- virtual world
- moving object trajectories
- model free
- globally optimal
- virtual reality
- machine learning
- learning process
- moving objects
- motion patterns
- state space
- case study
- real world
- function approximators
- consistency constraints
- markov decision process
- temporal difference
- action selection
- learning algorithm
- multi agent
- spatio temporal
- database
- dynamic programming
- dynamic environments