Haply Robotics and NVIDIA Introduce Touch-Enabled Physical AI Training at CES 2026
Haply and NVIDIA built a joint demo integrating force feedback into NVIDIA Isaac Sim — realistic physical interaction during data collection for more robust robotic AI models.
Originally published on haply.co — the canonical version of this article.
Haptics meets Isaac Sim
Haply Robotics and NVIDIA jointly developed a demonstration that integrates haptic force feedback directly into NVIDIA Isaac Sim, NVIDIA’s robotics simulation platform. Operators collecting training demonstrations feel the simulated environment as they work — contact, resistance, and weight — instead of steering blind through a purely visual scene.
Better data in, better models out
The premise is simple: realistic physical interaction during data collection yields more robust AI models. When the human demonstrator can feel what the robot would feel, demonstrations are more precise and more natural, and the recorded force-and-motion data is far richer training signal for contact-heavy tasks.
The full NVIDIA stack
The demo is enhanced with NVIDIA Cosmos for synthetic data generation — multiplying each human demonstration into many varied training examples — and runs on RTX 6000 Blackwell GPUs for real-time physics and rendering at haptic rates.
See it at CES 2026
The joint demonstration debuted at CES 2026 in Las Vegas, marking the deepest integration yet between Haply’s force-feedback hardware and NVIDIA’s Physical AI platform.