Afferent-Omni — Causal Vision-to-Touch Prior for Physical AI
Afferent-Omni is a closed-loop 50Hz haptic telemetry and 3D simulation platform designed as a causal vision-to-touch prior for Physical AI and robotic manipulation. Powered by NVIDIA Nemotron 3 Super 120B and FastAPI, it bridges visual perception with real-time tactile and proprioceptive feedback loops. Key Engineering Highlights: 50Hz Closed-Loop Haptic Telemetry: Low-latency tactile and proprioceptive signal streaming for deterministic physical interaction and contact modeling. Causal Vision-to-Touch Inference: Integrates NVIDIA Nemotron 3 Super 120B reasoning with 3D simulation environments to predict physical contact dynamics before actuation. High-Throughput Simulation Backbone: Built on asynchronous Python / FastAPI for real-time telemetry ingestion, state synchronization, and reproducible robotics benchmarking. Developed for the Nebius x NVIDIA Global AI Hackathon.
More from members
All projectsSee what other members are building and what each project is looking for.
Comments · 0