MIT's HardFlow enforces hard constraints on generative AI without retraining
MIT News AI · rss · 2026-09-14
MIT researchers led by Navid Azizan have developed HardFlow, a plug-and-play algorithm (published in IEEE TPAMI) that lets pretrained generative models satisfy hard safety, physical, and task-specific constraints at output time without sacrificing quality.
Key idea
- Conventional projection-based sampling forces constraints at every intermediate generation step, limiting exploration and degrading solution quality.
- HardFlow enforces constraints only on the final output, reformulating constrained sampling as a trajectory-optimization problem solved with optimal-control tools. Leveraging the structure of flow-matching models, it decomposes the problem into efficient single-step subproblems solvable at deployment time, with no retraining needed.
Results
- Across robotic manipulation, maze navigation, and text-guided image editing, HardFlow achieved 100% constraint satisfaction while finding better solutions than baselines — e.g., collision-free robotic paths that are also faster to the target, where other methods collided or took far longer.
- Runtime was comparable to or lower than competing methods, making it practical for safety-critical deployment of existing pretrained models.
More from Embodied
- Open-source companion robot Lamp starts shipping to 20+ countries at $969 — dee_hw · 2026-10-30
- SCAMPER demo: virtual rat trots and gallops with gaits it never saw in training — tyrell_turing · 2026-10-30
- DIY humanoid build, day 7: creator gives up fixing physics, adds 1-DOF ankles — _Stocko_ · 2026-10-30
- Lambda Surgical launches first-in-human study of LiDAR-based LambdaGuide at PINE — ditzikow · 2026-10-30
- Gaussian splats double as shared GPS: one city map localizes drones, robots and phones — bilawalsidhu · 2026-10-30
- SimReady dataset offers 2,677 simulation-ready 3D assets to build virtual test worlds — victormustar · 2026-10-30