mlsanityData to 7 Oct 2026

Parallel Differentiable Reachability for Learning and Planning with Certified Neural Dynamics and Controllers

arXiv 2605.25346 ↗First posted 25 May 2026cs.RO cs.AI cs.LGcited by 2

Not on the map. Why?

Abstract (arXiv)

Neural network (NN) dynamics models and control policies achieve strong performance in robotics, but providing sound guarantees under uncertainty remains difficult, especially for closed-loop NN systems. Existing reachability tools provide formal over-approximations, yet are often non-differentiable, overly conservative, or too slow for modern learning and online planning pipelines. To address this, we present a parallelizable, differentiable reachability framework in JAX for continuous- and discrete-time systems with analytical and NN-based dynamics and controllers. Our framework combines Taylor-model flowpipe construction with CROWN-style linear bound propagation through a unified representation that preserves affine dependencies while supporting GPU-batched computation and automatic differentiation. Building on this reachability primitive, we develop (i) a certified training method that encourages reachability-friendly dynamics models and controllers, and (ii) a reachability-aware sampling-based MPC scheme with gradient-based refinement. Experiments on non-prehensile manipulation and quadrotor tasks, including hardware and higher-dimensional evaluations (up to 72D), demonstrate practical online planning while maintaining certified reachable-set over-approximations under bounded uncertainty.

Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 48 of its 70 references that are in the map. First posted on arXiv 25 May 2026.

Builds on

48 of 70 references.

References · 48

  1. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 2,730
  2. 2019Dream to Control: Learning Behaviors by Latent Imaginationcited by 1,598
  3. 2015End-to-End Training of Deep Visuomotor Policiescited by 847
  4. 2017Provable defenses against adversarial examples via the convex outer adversarial polytopecited by 564
  5. 2018Efficient Neural Network Robustness Certification with General Activation Functionscited by 439
  6. 2022DayDreamer: World Models for Physical Robot Learningcited by 372
  7. 2018On the Effectiveness of Interval Bound Propagation for Training Verifiably Robust Modelscited by 322
  8. 2022On Neural Differential Equationscited by 264
  9. 2016Unsupervised Learning for Physical Interaction through Video Predictioncited by 256
  10. 2019Algorithms for Verifying Deep Neural Networkscited by 221
  11. 2018Visual Foresight: Model-Based Deep Reinforcement Learning for Vision-Based Robotic Controlcited by 220
  12. 2018Learning Particle Dynamics for Manipulating Rigid Bodies, Deformable Objects, and Fluidscited by 219
36 more
  1. 2019Towards Stable and Efficient Training of Verifiably Robust Neural Networkscited by 204
  2. 2020Fast and Complete: Enabling Complete Neural Network Verification with Rapid and Massively Parallel Incomplete Verifierscited by 183
  3. 2018Efficient Formal Safety Analysis of Neural Networkscited by 165
  4. 2020NNV: The Neural Network Verification Tool for Deep Neural Networks and Learning-Enabled Cyber-Physical Systemscited by 128
  5. 2019A Convex Relaxation Barrier to Tight Robustness Verification of Neural Networkscited by 124
  6. 2022General Cutting Planes for Bound-Propagation-Based Neural Network Verificationcited by 102
  7. 2020Keypoints into the Future: Self-Supervised Correspondence in Model-Based Reinforcement Learningcited by 69
  8. 2022RoboCraft: Learning to See, Simulate, and Shape Elasto-Plastic Objects with Graph Networkscited by 69
  9. 2021Training Certifiably Robust Neural Networks with Efficient Local Lipschitz Boundscited by 64
  10. 2019Neural Network Branching for Neural Network Verificationcited by 45
  11. 2024The Fifth International Verification of Neural Networks Competition (VNN-COMP 2024): Summary and Resultscited by 42
  12. 2023How to Train Your Neural Control Barrier Function: Learning Safety Filters for Complex Input-Constrained Systemscited by 40
  13. 2021POLAR: A Polynomial Arithmetic Framework for Verifying Neural-Network Controlled Systemscited by 38
  14. 2016Funnel Libraries for Real-Time Robust Feedback Motion Planningcited by 28
  15. 2019JuliaReach: a Toolbox for Set-Based Reachabilitycited by 28
  16. 2022Safe Control Under Input Limits with Neural Control Barrier Functionscited by 27
  17. 2025The 6th International Verification of Neural Networks Competition (VNN-COMP 2025): Summary and Resultscited by 26
  18. 2020Planning with Learned Dynamics: Probabilistic Guarantees on Safety and Reachability via Lipschitz Constantscited by 25
  19. 2024Lyapunov-stable Neural Control for State and Output Feedback: A Novel Formulationcited by 23
  20. 2023Dynamic-Resolution Model Learning for Object Pile Manipulationcited by 22
  21. 2017FaSTrack: a Modular Framework for Fast and Guaranteed Safe Motion Planningcited by 21
  22. 2022Safe Output Feedback Motion Planning from Images via Learned Perception Modules and Contraction Theorycited by 18
  23. 2023Exact Verification of ReLU Neural Control Barrier Functionscited by 14
  24. 2021Model Error Propagation via Learned Contraction Metrics for Safe Feedback Motion Planning of Unknown Systemscited by 13
  25. 2023POLAR-Express: Efficient and Precise Formal Reachability Analysis of Neural-Network Controlled Systemscited by 12
  26. 2022Statistical Safety and Robustness Guarantees for Feedback Motion Planning of Unknown Underactuated Stochastic Systemscited by 11
  27. 2018Robust Tracking with Model Mismatch for Fast and Safe Planning: an SOS Optimization Approachcited by 8
  28. 2024Verified Safe Reinforcement Learning for Neural Network Dynamic Modelscited by 7
  29. 2026Safety Beyond the Training Data: Robust Out-of-Distribution MPC via Conformalized System Level Synthesiscited by 7
  30. 2025Two-Stage Learning of Stabilizing Neural Controllers via Zubov Sampling and Iterative Domain Expansioncited by 6
  31. 2024BaB-ND: Long-Horizon Motion Planning with Branch-and-Bound and Neural Dynamicscited by 5
  32. 2025Neural Contraction Metrics with Formal Guarantees for Discrete-Time Nonlinear Dynamical Systemscited by 5
  33. 2024textttimmrax: A Parallelizable and Differentiable Toolbox for Interval Analysis and Mixed Monotone Reachability in JAXcited by 4
  34. 2024Certified Training with Branch-and-Bound for Lyapunov-stable Neural Controlcited by 3
  35. 2025Safe Domains of Attraction for Discrete-Time Nonlinear Systems: Characterization and Verifiable Neural Network Estimationcited by 3
  36. 2019Optimization Based Planner Tracker Design for Safety Guaranteescited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026Bridging Control with Neural Network Verifier alpha-beta-CROWN: A Tutorialcited by 1
  2. 2026GPU-Parallel Linearization Error Bounds for Real-Time Robust Optimal Control of Nonlinear and Neural Network Dynamicscited by 0