Adaptive reinforcement learning with active state-specific exploration for engagement maximization during simulated child-robot interaction.
George VelentzasTheodore TsitsimisIñaki RañóCostas S. TzafestasMehdi KhamassiPublished in: Paladyn J. Behav. Robotics (2018)
Keyphrases
- reinforcement learning
- real robot
- state space
- autonomous learning
- mobile robot
- human robot interaction
- human robot
- agent learns
- action selection
- vision system
- simulated robot
- robot control
- markov decision process
- active exploration
- perceptual aliasing
- robot manipulators
- control policy
- machine learning
- humanoid robot
- transition model
- initially unknown
- user interaction
- dynamic environments
- learning process
- learning environment