Physics-Based Dexterous Manipulations with Estimated Hand Poses and Residual Reinforcement Learning.
Guillermo Garcia-HernandoEdward JohnsTae-Kyun KimPublished in: CoRR (2020)
Keyphrases
- reinforcement learning
- human hand
- residual error
- function approximation
- markov decision processes
- learning algorithm
- learning process
- state space
- deformable models
- object manipulation
- optimal control
- multi agent
- data sets
- real time
- optimal policy
- multi modal
- video sequences
- estimation error
- temporal difference
- robot control
- finite element analysis
- neural network
- haptic feedback