Jim Fan trains robots on human video, calls VR teleop rigs unscalable 'torture devices'
dawnsongtweets · x · 2026-09-23
In a Berkeley RDI interview, Nvidia's Jim Fan slammed VR teleoperation rigs for robot training as "medieval torture devices" that fundamentally don't scale. He trains almost entirely on ordinary human video instead. The core debate: can video alone teach dexterity? You can see a hand move, but not the force it applied — a tradeoff between scale and missing force/haptic information.
More from Embodied
- RealSense launches people-detection challenge, best project wins a D421 camera — chrismatthieu · 2026-09-23
- IROS 2026 Hits Pittsburgh: 4,000+ Attendees, Nearly 2,000 Papers — heatherknight · 2026-09-23
- Dev builds an ESP32 face for his AI agent using Copilot CLI, open-sourced — DanWahlin · 2026-09-23
- seeMote Cube debuts: a spatial accessory for Vision Pro with IR 6DoF tracking — OwariDa · 2026-09-22
- Estimating Unitree G1 actuator heat loss in Isaac Sim with a physics-based model — IsaiahBallah · 2026-09-22
- RoboDawn: Tsinghua & Tencent Hunyuan drive robots with frozen VLM, 73.6% one-shot success — _akhaliq · 2026-09-22