At Nvidia GTC 2026, Techman Robot demonstrated a humanoid system that uses motion capture to train its TM Xplore I platform, highlighting a shift toward data-driven robot training. The company also announced a collaboration with j-mex, integrating wearable motion capture technology into its development workflow.
Key takeaways
- Operators wearing a VR headset and j-mex's Moxi motion capture suit can control the TM Xplore I humanoid in real time, replicating actions like dual-arm grasping and object sorting.
- The approach captures human motion to generate high-quality training data, supporting faster development of sim-to-real capabilities for dynamic physical environments.
- j-mex's IMU-based technology precisely tracks human movement, converting subtle actions into robotic training data.
- Techman Robot emphasizes integrating perception, decision-making, and motion control across collaborative and humanoid platforms to enable operation in less structured settings.
This demonstration underscores a broader industry move toward physical AI, where robots learn from real-world data rather than traditional programming. As Scott Huang, COO of Techman Robot, noted, future automation will advance toward deep, intelligent collaboration with perception and thinking capabilities.
Source: Robotics & Automation News (roboticsandautomationnews.com) · Published 2026-03-20 · “Techman Robot showcases humanoid system with motion capture training at GTC 2026”