Avoiding moving obstacles with stochastic hybrid dynamics using PEARL: PrEference Appraisal Reinforcement Learning.
Aleksandra FaustHao-Tien ChiangNathanael RackleyLydia TapiaPublished in: ICRA (2016)
Keyphrases
- reinforcement learning
- direct policy search
- stochastic approximation
- learning automata
- function approximation
- dynamic model
- optimal control
- markov decision processes
- learning process
- multi agent
- multi attribute
- multi criteria
- machine learning
- ground surface
- learning algorithm
- real estate
- dynamic programming
- moving objects
- real time
- optimal policy
- recommender systems
- conditional independence
- model free
- temporal difference
- dynamical systems
- monte carlo
- preference elicitation
- reinforcement learning methods
- user preferences
- state space