mlsanityData to 14 Oct 2026

Diffusion-Based Failure Sampling for Evaluating Safety-Critical Autonomous Systems

arXiv 2406.14761 ↗First posted 20 Jun 2024cs.RO cs.AI cs.SYcited by 6

On the map: Diffusion models → Diffusion and flow policies for robotics

Abstract (arXiv)

Validating safety-critical autonomous systems in high-dimensional domains such as robotics presents a significant challenge. Existing black-box approaches based on Markov chain Monte Carlo may require an enormous number of samples, while methods based on importance sampling often rely on simple parametric families that may struggle to represent the distribution over failures. We propose to sample the distribution over failures using a conditional denoising diffusion model, which has shown success in complex high-dimensional problems such as robotic task planning. We iteratively train a diffusion model to produce state trajectories closer to failure. We demonstrate the effectiveness of our approach on high-dimensional robotic validation tasks, improving sample efficiency and mode coverage compared to existing black-box techniques.

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Placed in Diffusion and flow policies for robotics, one of 14 directions in Diffusion models. Builds on 19 of its 35 references that are in the map. First posted on arXiv 20 Jun 2024.

Builds on

19 of 35 references.

Closest ancestors · 2

Same direction, newest first

  1. 2023DiffAIL: Diffusion Adversarial Imitation Learning11 Dec
  2. 2023Motion Planning Diffusion: Learning and Planning of Robot Motions with Diffusion Models3 Aug

Other references · 17

  1. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  2. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 9,122
  3. 2021Improved Denoising Diffusion Probabilistic Modelscited by 3,761
  4. 2018UMAP: Uniform Manifold Approximation and Projection for Dimension Reductioncited by 2,898
  5. 2022Planning with Diffusion for Flexible Behavior Synthesiscited by 1,061
  6. 2020Reliable Fidelity and Diversity Metrics for Generative Modelscited by 434
  7. 2019Self-Driving Cars: A Surveycited by 148
  8. 2018Scalable End-to-End Autonomous Vehicle Testing via Rare-event Simulationcited by 77
  9. 2020A Survey of Algorithms for Black-Box Safety Validation of Cyber-Physical Systemscited by 71
  10. 2020Neural Bridge Sampling for Evaluating Safety-Critical Autonomous Systemscited by 35
  11. 2019Efficient Black-box Assessment of Autonomous Vehicle Safetycited by 23
  12. 2018Falsification of Cyber-Physical Systems Using Deep Reinforcement Learningcited by 17
5 more
  1. 2023ShiftDDPMs: Exploring Conditional Diffusion Models by Shifting Diffusion Trajectoriescited by 14
  2. 2023A Flow-Based Generative Model for Rare-Event Simulationcited by 5
  3. 2023Model-based Validation as Probabilistic Inferencecited by 3
  4. 2023Requirement falsification for cyber-physical systems using generative modelscited by 3
  5. 2020Transfer Learning for Efficient Iterative Safety Validationcited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 3; 2026: 3.

6 papers.

  1. 2025Predictive Uncertainty for Runtime Assurance of a Real-Time Computer Vision-Based Landing Systemcited by 3
  2. 2025Diffusion Models for Safety Validation of Autonomous Driving Systemscited by 1
  3. 2026The FABRIC Strategy for Verifying Neural Feedback Systemscited by 1
  4. 2025Robust Planning for Autonomous Vehicles with Diffusion-Based Failure Samplerscited by 0
  5. 2026A New Strategy for Verifying Reach-Avoid Specifications in Neural Feedback Systemscited by 0
  6. 2026Importance Sampling and PCA for Finding Failures in Commercial Autonomous Vehiclescited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Diffusion-Guided Search via Exponential Tilting (DiffTilt): An Application to Falsification of Safety-Critical Systems
  2. 2026ROBOGATE: Adaptive Failure Discovery for Safe Robot Policy Deployment via Two-Stage Boundary-Focused Sampling ↗
  3. 2023SafeDiffuser: Safe Planning with Diffusion Probabilistic Models
  4. 2025From Demonstrations to Safe Deployment: Path-Consistent Safety Filtering for Diffusion Policies
  5. 2023Trajectory Generation, Control, and Safety with Denoising Diffusion Probabilistic Models
  6. 2024Denoising Diffusion Planner: Learning Complex Paths from Low-Quality Demonstrations
  7. 2025Model-Based Diffusion Sampling for Predictive Control in Offline Decision Making
  8. 2025Diffusion-SAFE: Diffusion-Native Human-to-Robot Driving Handover for Shared Autonomy