mlsanityData to 7 Oct 2026

Learning-Based Shielding for Safe Autonomy under Unknown Dynamics

arXiv 2410.07359 ↗First posted 7 Oct 2024eess.SY cs.LG cs.SYcited by 3

Not on the map. Why?

Abstract (arXiv)

Shielding is a common method used to guarantee the safety of a system under a black-box controller, such as a neural network controller from deep reinforcement learning (DRL), with simpler, verified controllers. Existing shielding methods rely on formal verification through Markov Decision Processes (MDPs), assuming either known or finite-state models, which limits their applicability to DRL settings with unknown, continuous-state systems. This paper addresses these limitations by proposing a data-driven shielding methodology that guarantees safety for unknown systems under black-box controllers. The approach leverages Deep Kernel Learning to model the systems' one-step evolution with uncertainty quantification and constructs a finite-state abstraction as an Interval MDP (IMDP). By focusing on safety properties expressed in safe linear temporal logic (safe LTL), we develop an algorithm that computes the maximally permissive set of safe policies on the IMDP, ensuring avoidance of unsafe states. The algorithms soundness and computational complexity are demonstrated through theoretical proofs and experiments on nonlinear systems, including a high-dimensional autonomous spacecraft scenario.

Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Builds on 14 of its 28 references that are in the map. First posted on arXiv 7 Oct 2024.

Builds on

14 of 28 references.

References · 14

  1. 2017Safe Reinforcement Learning via Shieldingcited by 427
  2. 2017On Kernelized Multi-armed Banditscited by 348
  3. 2021The Promises and Pitfalls of Deep Kernel Learningcited by 81
  4. 2015Shield Synthesis: Runtime Enforcement for Reactive Systemscited by 41
  5. 2023Safe Reinforcement Learning via Probabilistic Logic Shieldscited by 41
  6. 2019Formal Synthesis of Stochastic Systems via Control Barrier Certificatescited by 21
  7. 2019A Barrier Function Approach to Finite-Time Stochastic System Verification and Controlcited by 14
  8. 2022Formal Control Synthesis for Stochastic Neural Network Dynamic Modelscited by 7
  9. 2024Error Bounds For Gaussian Process Regression Under Bounded Support Noise With Applications To Safety Certificationcited by 6
  10. 2021Formal Verification of Unknown Dynamical Systems via Gaussian Process Regressioncited by 5
  11. 2023Promises of Deep Kernel Learning for Control Synthesiscited by 4
  12. 2024Data-Driven Permissible Safe Control with Barrier Certificatescited by 3
2 more
  1. 2023Formal Abstraction of General Stochastic Systems via Noise Partitioningcited by 2
  2. 2024Shielded Deep Reinforcement Learning for Complex Spacecraft Taskingcited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026Robust Shielding for Safe Reinforcement Learningcited by 2
  2. 2026Kernel-Based Safe Exploration in Deep Reinforcement Learningcited by 0
  3. 2026Adaptive Probabilistic Shielding by Learning MDPs for Safe Reinforcement Learningcited by 0