Mecka AI Raises $60M to Advance Human Motion Data for Robot Training

Image: TechCrunch · Source

Mecka AI secures $60 million Series B led by Sequoia to collect human motion data for training humanoid and other robots, aiming to support robotic learning through real-world task recordings.

Mecka AI, a startup founded in 2024, announced it has raised $60 million in a Series B funding round led by Sequoia Capital. Participation also came from Nvidia, Microsoft’s venture fund M12, and other investors, confirming the company's rapid rise in the robotics data sector at a valuation near $500 million, as previously reported by TechCrunch.

The company specializes in collecting and analyzing human motion data to enhance the training of humanoid robots and various robotic systems. To gather this data, Mecka AI incentivizes people to record themselves performing routine activities, such as making coffee or repairing cars. They do so while equipped with body sensors and smartphones, generating richly detailed datasets of human movement.

Mecka AI aims to be to robotics what data-labeling firms like Scale AI, Mercor, and Surge have become for large language models. These companies provide the essential human annotations that allow AI systems to learn and improve. By supplying comprehensive human motion datasets, Mecka AI supports the development of robots with enhanced physical task execution capabilities.

The robotics training data domain is attracting significant attention, with other startups like XDOF pursuing similar objectives and reportedly approaching valuations over $1 billion. Additionally, companies that historically focused on human data for language AI, such as Scale AI and Micro1, are expanding into robotics, highlighting the interdisciplinary nature of advanced AI training data needs.

As robots evolve toward greater autonomy and efficiency, access to varied and high-quality human motion data becomes increasingly critical. Mecka AI’s approach of paying individuals to record themselves in natural settings offers a scalable method to collect the diverse data necessary for training adaptable robotic systems. The latest funding round provides the resources to accelerate these efforts and broaden the company’s impact in the robotics field.