The Humyn Physical AI Data Engine is a data-collection and labeling platform by Humyn Labs that captures synchronized first-person stereo video, IMU telemetry, and audio, then processes and delivers time-aligned labeled datasets in standard robotics formats including MCAP, RLDS, and ROS Bag. Multi-stage quality checks—including human-in-the-loop validation—run at every stage of the pipeline.
Primary use case: Collecting synchronized, quality-controlled ego-centric training data for Physical AI and robotics.
AI Capabilities
1920×1080 RGB stereo first-person video capture at 30 fps, 300 Hz 6-axis IMU telemetry capture, 1 μs hardware synchronization between the stereo cameras and IMU, On-device recording with wireless upload, In-capture quality control that flags defective footage before collection ends, On-device consent confirmation, Optional mono or stereo microphone audio capture, USB power-bank operation
Primary Use Case
Collecting synchronized, quality-controlled ego-centric training data for Physical AI and robotics.
Physical AI Startup Humyn Hits $50 Million ARR In 6 Months, Raises $47 Million
A Bengaluru startup founded in February 2026 crossed a $50 million annualized revenue run rate in six months by supplying US robotics companies with the one type of data they cannot generate themselves: real humans doing real jobs in real places. Its Series A has an unusual structure — $26M in equity, $21M in venture debt — and the reason behind that split is a massive indicators on how the physical AI data market actually works.
Humyn Labs funding reaches $47 million in equity and debt to expand physical AI training data and robotics deployment. Explore India’s role.