Nomadic raises $8.4M to convert drone video into training datasets

Nomadic has raised $8.4 million in seed funding to solve one of the critical infrastructure problems in the autonomous transport industry: what to do with the massive volumes of video data continuously generated by autonomous vehicles and robots. The company has developed a platform based on deep learning that automatically converts raw video streams from autonomous system cameras into structured, searchable datasets. Instead of spending months on manual annotation of terabytes of recordings, developers of robots and autonomous vehicles get ready-to-use annotated databases.

The scale of the problem Nomadic solves is easy to underestimate. A single autonomous vehicle equipped with a standard sensor suite—cameras, lidars, radars—can generate between one and several terabytes of data per hour. Multiply that by a fleet of hundreds or thousands of vehicles operating in test and commercial programs worldwide, and the scale of the challenge becomes clear: we're talking about petabytes of video material that need to be stored, processed, and made suitable for training the next generation of models.

The traditional approach—manual data annotation through outsourcing platforms—works, but is slow and expensive. Nomadic's new approach: automate the entire process using its own deep learning model, trained to understand context in automotive and robotics environments.

The practical result: engineers can not only store video recordings, but also work with them as a full-fledged database. You can submit a query like "find all scenes where a pedestrian steps onto the road in poor lighting conditions" or "show cases where the vehicle encountered an unrecognized obstacle"—and get relevant episodes in seconds, rather than weeks of manual review.

The $8.4 million investment round looks modest compared to billion-dollar investments in autonomous system development itself, but reflects growing market understanding: data infrastructure is as critical a component as the perception algorithms themselves. Without effective management of training data, autonomous technology development hits an operational ceiling.

Nomadic operates in a niche that is becoming increasingly competitive as the autonomous systems market matures. Interest in ML data management tools surged after major players publicly acknowledged that training data quality is one of their key competitive advantages. The category of tools for processing and annotating data in the autonomous systems field is emerging from the shadows. If previously such companies remained invisible suppliers to major players, today investors are beginning to see independent value in them—analogous to how developer tools became a separate multi-billion-dollar SaaS segment.

Nomadic is betting that the winner in this race will be whoever first gives autonomous vehicle developers an answer to the question: what do we do with all this video.