Innodata opens motion-capture lab for robotics training data
Innodata Inc. (NASDAQ: INOD) has opened a motion-capture research and development laboratory in New Jersey designed to generate training data for humanoid robots and other physical AI systems, according to a company statement.
The lab was developed in collaboration with Vicon, a motion-capture technology company that consulted on the lab's design. The facility uses infrared optical tracking cameras capable of measuring movement at the sub-millimeter level, which Innodata says offers greater precision than wearable inertial measurement unit sensors or single-camera video analysis.
Unlike data providers that infer three-dimensional motion from two-dimensional video, Innodata says its lab captures 3D data directly from human or mechanical subjects.
"There's just no substitute for direct 3D motion capture," said Frank Tanner, vice president of robotics and physical AI at Innodata. "When you're training a humanoid that weighs almost 200 pounds, your readings can't be in the ballpark. They need to be precise."

Andrew Knox, managing director of Vicon, said the system captures people, robots, and objects in the same space with sub-millimeter accuracy and millisecond latency, providing what he described as "independent ground truth, not an estimate."
Customers can purchase motion-capture data as off-the-shelf packages or commission custom projects. The lab also offers robot evaluation services, using its external camera systems to verify a robot's movements against its internal telemetry data across scripted scenarios, including human-robot interaction.
Innodata CEO Rahul Singhal said the facility gives physical AI companies "a complete end-to-end capability — from data collection through model evaluation."
The lab is part of Innodata's broader physical AI practice, which covers data capture across humanoids, wearable systems, sensor rigs, and Universal Manipulator Interface grippers, as well as safety-assurance services.

