Ropedia, a Singapore-based startup, has raised $30 million in a Pre-A round. These funds are intended to expand its data collection platform for real-world interaction, which is critical for the development of robotics and physical artificial intelligence.
The Challenge of Training Physical AI
Training AI to interact with the physical world is one of the most complex challenges of our time. Unlike chatbots that learn from internet text content, robots require practical experience. Ropedia addresses this problem by creating a platform that allows robots to master complex tasks such as electronics assembly or tool handling.
Data Collection Methodology
The company was founded in late 2025 by Zaoxi Chen, Fangzhou Hong, and Zhiwei Liu. The company's headquarters are located in Singapore and Mountain View, California. The founders leverage their extensive research knowledge gained at Meta, Nanyang Technological University, and in 3D computer vision to solve a key problem in training physical AI.
Traditional methods of training AI based on collecting data from public websites are insufficient for physical AI. Robots need to receive actual sensory data synchronized with human movements, including the interpretation of tactile pressure, camera angles, body position, and timing. Ropedia generates this raw data directly, rather than simply relabeling existing datasets.
Technological Advantage
The company's solution is 50 times cheaper than a classic teleoperation setup for data collection. Instead of expensive equipment for every experiment, human collectors gather data in ordinary conditions. The company's main dataset is called Xperience-10M, which already contains over 10,000 hours of multimodal recordings and 10 million interaction episodes, providing robotics laboratories with structured logs of real human actions for training.
HOMIE Device and Synchronization
At the core of Ropedia's data engine is HOMIE—a proprietary head-mounted device that simultaneously captures multimodal streams. Furthermore, HOMIE records video at frame rates, audio, depth maps, hand tracking, gaze direction, body movement, and camera pose. A critically important aspect is that all data streams in HOMIE have precise timestamps, as physical AI models require the system to perceive, feel, and act identically every millisecond. HOMIE eliminates the synchronization problem present in most existing tools.
Due to the lightweight and portable nature of the wearable device, it can be used by anyone on a factory floor, in a kitchen, or in a repair shop. Ropedia has already begun mass production of HOMIE for delivery to technology partners and commercial clients.
Market Expansion and Plans
The $30 million capital was secured in two Pre-A tranches. In mid-March 2026, the company closed a strategic seed funding round of $8 million and a new round of $22 million. The round was supported by private investors, renowned AI researchers, and partners from the corporate technology and robotics sectors.
The new capital will be directed towards expanding Ropedia's data collection fleets in Southeast Asia and North America. Additionally, the startup will increase the deployment pace of HOMIE to capture a wider range of 'real' behavior. A portion of the funds will also go toward strategic hiring, especially engineers at the Mountain View office. Ropedia's data infrastructure already supports more than a dozen North American companies developing foundational models for autonomous systems, providing robotics developers with the necessary foundation to create smarter and more capable machines.
