Ex-Meta scientists want to bring visual AI to the factory floor
techcrunch ·
Perceptron, a startup founded by two former Meta research scientists, has launched Isaac 0.5, an open-weight visual AI model designed to help robots operate in industrial settings like warehouses and factory floors. The company was founded in November 2024 by Armen Aghajanyan and Akshat Shrivastava, both previously with Meta's Fundamental AI Research division, and recently raised $21 million led by Bessemer Venture Partners. Isaac 0.5 aims to give machines the ability to perceive, reason, and act in physical environments, while also extracting visual intelligence from videos recorded by robots. Unlike narrow task-specific models, it is positioned as a general-purpose system that can flexibly handle multiple steps in complex processes such as package sorting. The model was trained on large amounts of video data, including general video, ego video, and UMI video, with the company building petabyte-scale internal datasets. Perceptron plans to market the software across manufacturing, logistics, warehousing, security, mobility, and media industries, claiming no similar product currently exists.
techcrunch