ORN Introduction
ORN is the proprietary egocentric video data processing, evaluation & labeling infrastructure that powers EgoPlay. It transforms raw egocentric videos into structured, robotics policy training-ready datasets through a layered architecture. It is being developed by our infrastructure partner Monitize AI.
ORN turns raw, first-person video into privacy-safe, processed, qualified datapoints. Its layered pipeline - Ingestion for technical compliance & authenticity, Pre-Processing for video quality & privacy, Evaluation for understanding & skill-scoring, and Post-Processing for normalization - creates a dependable foundation for EgoPlay’s merit-based economy while producing robotics policy training-ready datasets enriched with both low-level action labels and high-level semantic narratives enabling both perception and policy learning.
In short, ORN is an engine that converts everyday human activity, recorded from an egocentric perspective, into structured training knowledge for Physical AI.
Orn’s architecture is divided into 4 core layers:
Data Ingestion Layer
Pre-Processing Layer
Evaluation Layer
Post-Processing Layer
Last updated