Human locomotion with reinforcement learning using bioinspired reward reshaping strategies.
Katharine NowakowskiPhilippe CarvalhoJean-Baptiste SixYann MailletAnh Tu NguyenIsmail SeghiriLoick M'PembaTheo MarcilleSy Toan NgoTien-Tuan DaoPublished in: Medical Biol. Eng. Comput. (2021)
Keyphrases
- reinforcement learning
- function approximation
- reward function
- human interaction
- eligibility traces
- exploration strategy
- policy gradient
- robot control
- biologically inspired
- transfer learning
- optimal policy
- machine learning
- state space
- learning algorithm
- multi agent
- markov decision processes
- human subjects
- optimal control
- mobile robot
- dynamic programming
- optimal strategy
- learning agents
- learning process
- neural network
- state action
- average reward
- learning agent
- temporal difference
- human users
- degrees of freedom
- supervised learning