Ropedia, a Singapore-based physical-AI startup spun out of Nanyang Technological University, has launched a program giving academic researchers access to its wearable data-capture device.
In a statement on Tuesday, Ropedia said the move can help build infrastructure for training robots and AI systems to operate in the real world.
Under the Academic Partner Program, universities and researchers can obtain a research kit built around HOMIE Gen2. As a head-mounted device, HOMIE Gen2 can record synchronized real-world data, such as video, motion, depth, spatial, inertial, and other signals, as people move and handle objects. The captured data can feed research into robotics, embodied AI and so-called world models and vision-language-action systems, Ropedia said.
Physical AI needs a different kind of training data than text-based generative AI, particularly records of real human behavior in the physical world, Ropedia said. It positions itself as a data-infrastructure layer for the field, handling capture, processing and model-ready datasets.
Nanyang Technological University, the source of Ropedia’s core technology, was among the first to trial the device, and researchers at its computing school continue to use it. Ropedia said its academic network also includes Princeton, Carnegie Mellon, the University of California San Diego, the University of Illinois Urbana-Champaign, Hong Kong Polytechnic University, Westlake University, and the Allen Institute for AI.
Co-founder and chief executive Zhaoxi Chen said the program can give researchers in Singapore and elsewhere access to real-world data infrastructure and to advance physical and embodied AI. Ropedia also said it was starting a 500-hour physical-AI evaluation program as it broadens its work.