RL + Model-based Control: Using On-demand Optimal Control to Learn Versatile Legged Locomotion.
Dongho KangJin ChengMiguel ZamoraFatemeh ZargarbashiStelian CorosPublished in: CoRR (2023)
Keyphrases
- optimal control
- control problems
- reinforcement learning
- control strategy
- dynamic programming
- legged locomotion
- feedback control
- infinite horizon
- actor critic
- control law
- model free
- rl algorithms
- control strategies
- linear quadratic
- rough terrain
- optimal control problems
- real time
- mathematical model
- reinforcement learning methods
- state space
- neural network
- function approximation
- policy gradient
- learning algorithm
- stochastic control