Continual Policy Distillation of Reinforcement Learning-based Controllers for Soft Robotic In-Hand Manipulation.
Lanpei LiEnrico DonatoVincenzo LornonacoEgidio FaloticoPublished in: RoboSoft (2024)
Keyphrases
- reinforcement learning
- optimal policy
- manipulation tasks
- policy search
- control policies
- action selection
- real robot
- markov decision process
- markov decision problems
- function approximators
- function approximation
- policy iteration
- markov decision processes
- partially observable markov decision processes
- partially observable
- policy evaluation
- model free
- control policy
- rl algorithms
- reinforcement learning problems
- partially observable environments
- reinforcement learning algorithms
- reward function
- state space
- mobile robot
- actor critic
- policy gradient
- approximate dynamic programming
- decision problems
- state action
- continuous state spaces
- machine learning
- action space
- partially observable domains
- real time
- dynamic programming
- robotic systems
- continuous state
- optimal control
- multi agent
- state and action spaces
- robotic control
- temporal difference
- hand gestures
- control loop
- inverse reinforcement learning
- learning algorithm
- motion planning
- control strategies
- control system
- transition model
- state dependent