Formally Verifying Deep Reinforcement Learning Controllers with Lyapunov Barrier Certificates.
Udayan MandalGuy AmirHaoze WuIeva DaukantasFletcher Lee NewellUmberto J. RavaioliBaoluo MengMichael DurlingMilan GanaiTobey ShimGuy KatzClark W. BarrettPublished in: CoRR (2024)
Keyphrases
- reinforcement learning
- adaptive control
- control law
- function approximation
- state space
- asymptotic stability
- closed loop
- dynamical systems
- model free
- control scheme
- optimal control
- class of nonlinear systems
- sufficient conditions
- reinforcement learning methods
- reinforcement learning algorithms
- controller design
- control theory
- markov decision processes
- control system
- stability analysis
- nonlinear systems
- lyapunov function
- policy search
- learning algorithm
- multi agent
- learning process
- machine learning
- transfer learning
- model checking
- optimal policy
- deep learning
- dynamic programming
- supervised learning
- chaotic systems
- behavioural cloning
- robust stability
- access control
- control strategies
- function approximators
- public key
- temporal difference
- control method