Continual Policy Distillation of Reinforcement Learning-based Controllers for Soft Robotic In-Hand Manipulation.
Lanpei LiEnrico DonatoVincenzo LomonacoEgidio FaloticoPublished in: CoRR (2024)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- manipulation tasks
- action selection
- control policies
- markov decision process
- state space
- partially observable
- policy iteration
- markov decision problems
- function approximation
- markov decision processes
- reinforcement learning algorithms
- reinforcement learning problems
- reward function
- action space
- control policy
- real robot
- function approximators
- partially observable markov decision processes
- policy evaluation
- approximate dynamic programming
- reinforcement learning methods
- rl algorithms
- average reward
- model free
- state action
- learning algorithm
- actor critic
- state and action spaces
- robot control
- machine learning
- continuous state spaces
- partially observable environments
- multi agent
- state dependent
- mobile robot
- policy gradient
- temporal difference
- infinite horizon
- optimal control
- robotic systems
- long run
- partially observable domains
- robotic control
- control loop
- object manipulation
- exploration exploitation tradeoff
- agent receives
- adaptive control
- agent learns
- inverse reinforcement learning
- learning process
- partially observable markov decision process
- robotic arm
- dynamic programming
- decision problems
- hand gestures