Dynamic Model Predictive Shielding for Provably Safe Reinforcement Learning.
Arko BanerjeeKia RahmaniJoydeep BiswasIsil DilligPublished in: CoRR (2024)
Keyphrases
- dynamic model
- reinforcement learning
- experimental data
- state space
- function approximation
- control scheme
- multiple models
- worst case
- learning algorithm
- optimal policy
- model free
- markov decision processes
- machine learning
- neural network
- real time
- shear stress
- unscented kalman filter
- parallel manipulator
- control strategies
- adaptive control
- numerical simulations