RACER: Epistemic Risk-Sensitive RL Enables Fast Driving with Fewer Crashes.
Kyle StachowiczSergey LevinePublished in: CoRR (2024)
Keyphrases
- risk sensitive
- model free
- markov decision processes
- optimal control
- reinforcement learning
- utility function
- control policies
- markov decision chains
- optimality criterion
- markov decision problems
- optimal policy
- reinforcement learning algorithms
- control strategies
- state space
- function approximation
- temporal difference
- expected utility
- finite state
- partially observable
- policy iteration
- action space
- control strategy
- planning problems
- markov chain