PTLD: Sim-to-Real Tactile Transfer with Minimal Real Data
JitendraMalikCV · x · 2026-07-13
This post shares a new paper on tactile policy training: PTLD: Sim-to-Real Privileged Tactile Latent Distillation for Dexterous Manipulation.
The author notes that previous sim-to-real work often relied heavily on zero-shot transfer, limiting them to weaker sensors (e.g., proprioception only). PTLD's approach is that by allowing the collection of a small amount of real-world data, it's possible to deploy stronger sensor-rich motor policies, achieving significantly better manipulation performance.
The paper is a collaboration between researchers from CMU, UW, UC Berkeley, and Meta FAIR, and the post includes links to the paper, project page, and a video.
Related event: PTLD Enables Tactile Dexterous Manipulation Without Tactile Simulation(3 posts)→
More from Embodied
- WanGP v18.81 adds Tencent's Hy-Nomas world model, playable on 8GB VRAM — cocktailpeanut · 2026-10-27
- NVIDIA open-sources FLORA, a robotics data engine ~125x cheaper than GPT-6 Astra — nvidia · 2026-10-27
- Embodied AI startups need a different operational discipline, investor says — ___Mufasaa · 2026-10-27
- π0.5 study: raising demos from 10 to 438 lifts Place Ball in Bowl success from 7% to 83% — stingning · 2026-10-27
- GPT-6 Astra controls dexterous hands zero-shot at just 5.2% success in 500 trials — stingning · 2026-10-27
- HandsOn benchmark: 66 verified tasks and 18,000 human demos for dexterous manipulation — stingning · 2026-10-27