IMLE-VLA replaces flow matching with single-step cIMLE, 3.67x faster robot actions at 55Hz
petitegeek · x · 2026-09-28
Researchers from SFU and UPenn introduced IMLE-VLA, tackling the slow, jerky inference of vision-language-action (VLA) policies.
- Key change: Existing VLAs like π0.5 use a 10-step flow-matching action head; IMLE-VLA swaps it for a single-step conditional IMLE (cIMLE) generator, keeping the same VLM backbone but requiring one forward pass instead of ten.
- Results: Inference jumps from 15Hz to 55Hz (3.67x), with up to 11x higher action throughput. It achieves the highest average success rate on the 40-task LIBERO benchmark (98.0%).
- Theory: cIMLE provably preserves multimodal action coverage, avoiding mode collapse without multi-step sampling.
- Real-world: Franka Panda experiments show 2.2–3.0x lower jerk and 3.9–6.6x lower VLA compute time, with smoother motion and much faster task completion.
Paper and code are open source; the work will be presented at IROS 2026.
More from Embodied
- YC-backed Ornadyne's flapping-wing drone was classified as a bird by Air Force detectors — import_jmr · 2026-11-16
- Uber starts offering Amazon's Zoox robotaxi rides in Las Vegas — Polymarket · 2026-11-16
- Ant Lingbo Embodied AI Challenge crowns TARS team after 48-hour real-robot hackathon — 新智元 · 2026-11-16
- Tesla FSD fleet tops 7 billion city miles, adding ~500M in under four weeks — XFreeze · 2026-11-16
- Ropedia Releases Xperience 2: 1,500+ Scanned Real-World Scenes Paired With Egocentric Demos — liuziwei7 · 2026-11-16
- Hangzhou's Manycore Tech launches Aholo MS07 3D laser scanner with CHCNAV — teortaxesTex · 2026-11-16