Human Archive raises $8.2M for robotics training data
human archive raised $8.2m to collect egocentric and tactile data from india's gig workers to train physical ai models.
human archive secured $8.2m from wing vc, nvp capital, and y combinator to build a dataset for physical ai by outfitting service workers with custom capture rigs. the startup faces friction from potential partners like urban company and pronto but maintains over 1,000 active headsets and 50 distinct sensor devices in the field. by synchronizing rgb-d video with tactile force and motion data, the company aims to provide the rich, real-world training data required for advanced robotics. this approach leverages india's scalable gig workforce to solve the bottleneck of high-quality physical demonstrations.