Learning to Locomote: Understanding How Environment Design Matters for Deep Reinforcement Learning.
Daniele RedaTianxin TaoMichiel van de PannePublished in: MIG (2020)
Keyphrases
- reinforcement learning
- learning process
- learning algorithm
- solving problems
- supervised learning
- deeper understanding
- learning problems
- online learning
- learning systems
- dynamic environments
- real time
- learning agent
- learning tasks
- temporal difference learning
- collaborative learning environment
- mobile robot
- autonomous learning
- learning agents
- longer term
- multi agent environments
- robot control
- simulated robot
- inquiry based learning
- learning capabilities
- action selection
- complex environments
- function approximation
- mobile learning
- design process
- unsupervised learning
- knowledge acquisition
- state space
- dynamic programming
- case study
- neural network