Verifiable Model-Free Safety Filters via Reinforcement Learning
On the map: Deep reinforcement learning methods → Safe and constrained RL
Abstract (arXiv)
This paper presents a reinforcement learning approach of a model-free safety filter, drawing inspiration from the framework of model-based Predictive Safety Filters (PSFs). Similar to conventional PSFs, our method adopts a Quadratic Programming (QP) formulation by representing the filter as an unrolled QP solver network. However, unlike existing PSFs that derive QP parameters explicitly from system models, we learn these parameters directly through Deep Reinforcement Learning (DRL), thereby eliminating the dependency on accurate system identification. Furthermore, compared to traditional neural network-based methods, this QP structure allows us to furnish a formal certificate for the persistent safety of the learned filter. Numerical results demonstrate that our method outperforms both conventional model-based PSFs and RL-trained Multi-Layer Perceptron (MLP) baselines in terms of safety guarantees, minimal intervention, and per-step computational load.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Safe and constrained RL, one of 15 directions in Deep reinforcement learning methods. Builds on 18 of its 35 references that are in the map. First posted on arXiv 7 May 2026.
Builds on
18 of 35 references.
Closest ancestors · 1
Same direction, newest first
Other references · 17
- 2017Proximal Policy Optimization Algorithmscited by 15,215
- 2019Algorithm Unrolling: Interpretable, Efficient Deep Learning for Signal and Image Processingcited by 421
- 2016Control Barrier Function Based Quadratic Programs for Safety Critical Systemscited by 287
- 2018A predictive safety filter for learning-based control of constrained nonlinear dynamical systemscited by 118
- 2021Safe Nonlinear Control Using Robust Neural Lyapunov-Barrier Functionscited by 94
- 2020Learning Control Barrier Functions from Expert Demonstrationscited by 92
- 2020DeepReach: A Deep Learning Approach to High-Dimensional Reachabilitycited by 91
- 2023The Safety Filter: A Unified View of Safety-Critical Control in Autonomous Systemscited by 88
- 2018Linear model predictive safety certification for learning-based controlcited by 37
- 2021Scalable Learning of Safety Guarantees for Autonomous Systems using Hamilton-Jacobi Reachabilitycited by 14
- 2022Self-Supervised Online Learning for Safety-Critical Control using Stereo Visioncited by 6
- 2024Sensor-Based Distributionally Robust Control for Safe Robot Navigation in Dynamic Environmentscited by 5
5 more
- 2023Linear Model Predictive Control under Continuous Path Constraints via Parallelized Primal-Dual Hybrid Gradient Algorithmcited by 3
- 2023MPC-Inspired Reinforcement Learning for Verifiable Model-Free Controlcited by 3
- 2023Constraint-Guided Online Data Selection for Scalable Data-Driven Safety Filters in Uncertain Robotic Systemscited by 2
- 2024Differentiable Predictive Control for Robotics: A Data-Driven Predictive Safety Filter Approachcited by 2
- 2025Control Barrier Function-Based Safety Filters: Characterization of Undesired Equilibria, Unbounded Trajectories, and Limit Cyclescited by 1
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2024Q-learning-based Model-free Safety Filter
- 2025Learning State-Action Control Barrier Functions for Model-Free Control under Constraints
- 2025Provably Optimal Reinforcement Learning under Safety Filtering
- 2026Safe Reinforcement Learning using Ideas from Model Predictive Control
- 2025CBF-RL: Safety Filtering Reinforcement Learning in Training with Control Barrier Functions
- 2025Explicit Control Barrier Function-based Safety Filters and their Resource-Aware Computation
- 2024Safety Filtering While Training: Improving the Performance and Sample Efficiency of Reinforcement Learning Agents
- 2025Safe model-based Reinforcement Learning via Model Predictive Control and Control Barrier Functions