Ropedia Launches HOMIE Gen2 to Scale Real-World Data for Physical AI
aiinsider ·
Ropedia, a Singapore-based company, has launched HOMIE Gen2, a wearable head-mounted data-capture system designed to record real-world human activity and convert it into training data for robots and physical AI. The 380-gram device captures 360-degree visuals with 1080p video at 30 frames per second using a global shutter, along with four-channel spatial audio, and supports synchronized capture across multiple units. Recordings are processed through Ropedia's Human Experience Engine, which structures over 10 data types such as video, audio, depth, movement, body pose, and spatial signals, with reported accuracy up to 96%. The launch follows a recent pre-Series A funding round that brought Ropedia's total funding to $30 million, and the device is the fourth generation of its capture hardware developed over the past year. The company's CEO noted that physical AI requires diverse, real-world data, which traditional fixed-camera setups cannot provide. HOMIE Gen2 is hands-free and self-contained, allowing data to be captured in natural environments like homes, factories, and retail locations. The system builds on Ropedia's Xperience-10M dataset, which includes over 10 million interaction episodes from 10,000 hours of first-person recordings.
aiinsider