Wutong Sensing: Building the tactile data layer for physical AI
创业邦 · wechat · 2026-08-25
Tactile sensing is a critical bottleneck in physical AI. This article profiles Wutong Sensing (悟通感控), a company building data infrastructure for physical AI via force-sensing gloves.
- The Problem: Vision alone cannot measure force, slip, or friction, which are crucial for delicate robot manipulation. Public tactile data is scarce.
- The Solution: Wutong's THD1000R glove uses proprietary "off-electric" flexible materials to capture 3D force vectors across the palm with high stability and low cost.
- Proven Track Record: Before robotics, the company successfully applied this methodology to battery safety, monitoring expansion force to predict thermal runaway 30-50 minutes early. They have accumulated 500TB of battery data.
- Business Model: Transitioning from hardware sales to a "data-as-a-service" model, supplying tactile data for robot training and safety monitoring for battery manufacturers.
More from Embodied
- LeCun: Hooking a humanoid robot to a data center LLM won't clean your room — _Stocko_ · 2026-08-25
- AR Visualization Cuts Teleoperation Time by 24% via Impedance Control — rsasaki0109 · 2026-08-25
- BrainCo's Dexterous Data Matrix combines human, robot, and virtual data for embodied AI — eyishazyer · 2026-08-25
- Europe is a lot less behind in robotics than people think — StephanSturges · 2026-08-25
- Robotic demos stuttering? FlashRT engine offers Hz-level latency benchmarks — Xianbao_QIAN · 2026-08-25
- X-Humanoid's Tiangong robot hits a 3.4m standing long jump on raw torque — CyberRobooo · 2026-08-25