Learning safe neural network controllers with barrier certificates
From MaRDI portal
Abstract: We provide a novel approach to synthesize controllers for nonlinear continuous dynamical systems with control against safety properties. The controllers are based on neural networks (NNs). To certify the safety property we utilize barrier functions, which are represented by NNs as well. We train the controller-NN and barrier-NN simultaneously, achieving a verification-in-the-loop synthesis. We provide a prototype tool nncontroller with a number of case studies. The experiment results confirm the feasibility and efficacy of our approach.
Recommendations
- Learning safe neural network controllers with barrier certificates
- An iterative scheme of safe reinforcement learning for nonlinear systems via barrier certificate generation
- Automated and formal synthesis of neural barrier certificates for dynamical models
- Generating probabilistic safety guarantees for neural network controllers
- Simulation based computation of certificates for safety of dynamical systems
Cited in
(18)- Training classifiers for feedback control with safety in mind
- Automated and formal synthesis of neural barrier certificates for dynamical models
- Sandboxing controllers for stochastic cyber-physical systems
- Neural flocking: MPC-based supervised learning of flocking controllers
- Verisig
- Gray-box adversarial testing for control systems with machine learning components
- Learning safe neural network controllers with barrier certificates
- Safe reinforcement learning: A control barrier function optimization approach
- Verifying Neural Network Controlled Systems Using Neural Networks
- Compositional synthesis of control barrier certificates for networks of stochastic systems against \(\omega\)-regular specifications
- Generating probabilistic safety guarantees for neural network controllers
- Run-time optimization for learned controllers through quantitative games
- A learner-verifier framework for neural network controllers and certificates of stochastic systems
- Hybrid controller synthesis for nonlinear systems subject to reach-avoid constraints
- Verification-guided programmatic controller synthesis
- Probabilistic reach-avoid for Bayesian neural networks
- Towards neural-network-guided program synthesis and verification
- An iterative scheme of safe reinforcement learning for nonlinear systems via barrier certificate generation
This page was built for publication: Learning safe neural network controllers with barrier certificates
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5918787)
