Offline Risk-sensitive RL with Partial Observability to Enhance Performance in Human-Robot Teaming.
Giorgio AngelottiCaroline P. C. ChanelAdam Henrique Moreira PintoChristophe LounisCorentin ChauffautNicolas DrougardPublished in: AAMAS (2024)
Keyphrases
- partial observability
- risk sensitive
- human robot
- reinforcement learning
- markov decision processes
- partially observable
- model free
- action selection
- optimal control
- markov decision problems
- markov decision process
- human robot interaction
- control policies
- partially observable markov decision processes
- dialogue system
- state space
- utility function
- optimal policy
- finite state
- policy iteration
- belief state
- function approximation
- robotic systems
- planning problems
- reinforcement learning algorithms
- learning algorithm
- dynamic programming
- average cost
- action space
- infinite horizon
- learning agent
- decision theoretic
- humanoid robot
- reward function
- temporal difference
- evolutionary computation
- knowledge acquisition