platoseedplatoseed
Multimodal data provider for robotics and world modeling
We’re archiving the physical world for embodied intelligence by collecting and labeling aligned multimodal data. To build dexterous and perceptive robots that generalize robustly, we need massive amounts of real-world data across multiple modalities and environments. We have thought deeply about the fine line between biomimicry and its application to humanoid systems. Based on this research, we design and deploy custom hardware across residential and manufacturing settings. We then post-process the resulting data through internal QA, anonymization, and annotation pipelines to deliver diverse, high-fidelity datasets at scale to frontier labs developing robotics foundation models and general-purpose robotics companies. We believe we are at a historic inflection point, with a unique opportunity to leave a dent on humanity and reshape physical labor markets forever. That's why our team dropped out of Stanford and Berkeley and moved to Asia to collect the world’s largest annotated multimodal dataset.
Human Archive is a multimodal data platform that collects, labels, and synchronizes video, sensor, audio, and long-horizon data from 100,000+ contributors and 500+ industry partners across homes, hotels, restaurants, agriculture, industrial, construction, and retail. The data is foundational infrastructure for training robots and understanding human embodied intelligence.
Human Archive operates as a data lab offering synchronized multimodal datasets: RGB, RGB-D video, audio, tactile sensor, and mocap data collected from real-world environments. Data is captured from contributors and partners across diverse domains and made available to researchers and robotics companies for training world models and embodied AI systems.
Who it’s for: Robotics companies, AI researchers, and automation engineers building systems that need real-world human motion and interaction data for training policies and understanding human behavior.
Recently raised $8.2M seed round; hiring for Head of Engineering, Embedded/Electrical Engineers, Firmware Engineer, Head of Operations, and Software Research Engineer; Stanford and Berkeley researcher team; Y Combinator backed.
Archiving the structure of human interaction in the physical world. Berkeley dropout and previous farmer (sold mangoes & planted trees)
building multimodal real-world datasets for robotics | prev. UC Berkeley MET (IEOR + Business)
Let's archive the world for embodied intelligence
Human Archive teams build infrastructure to collect, annotate, and provide multimodal sensorimotor data from real-world environments (homes, hospitality, retail, transportation, industrial, etc.) at scale. They announce HA-Multi and HA-Ego datasets with vision, tactile, depth, hand and pose data, plus annotations and SLAM, aimed at advancing embodied spatial intelligence for robotics and world-model evaluations, supported by a large contributor network and data pipelines.
Real-world human movement data for humanoid robots
Documenting human dexterity
