SMiRL: Surprise Minimizing RL in Dynamic Environments.
Glen BersethDaniel GengColine DevinChelsea FinnDinesh JayaramanSergey LevinePublished in: CoRR (2019)
Keyphrases
- dynamic environments
- reinforcement learning
- mobile robot
- autonomous agents
- reinforcement learning algorithms
- reinforcement learning agents
- path planning
- potential field
- collision avoidance
- autonomous systems
- highly dynamic environments
- learning agent
- single agent
- adaptive control
- markov decision processes
- agent systems
- real environment
- function approximation
- optimal policy
- state space
- action selection
- temporal difference
- model free
- changing environment
- belief space
- cooperative
- real time