Ex-Meta scientists want to bring visual AI to the factory floor

Two former Meta research scientists, Armen Aghajanyan and Akshat Shrivastava, have founded a startup called Perceptron with the goal of bringing visual artificial intelligence to the factory floor. The company launched its latest model, Isaac 0.5, this week, which aims to help machines perceive, reason, and act in industrial settings.

The software is capable of assisting vision-guided robots in navigating complex environments like warehouses or factory floors and extracting visual intelligence from videos recorded by those bots. Isaac 0.5 is released as an open-weight model, allowing anyone to inspect its parameters and training materials. The startup recently raised $21 million in funding, which was led by Bessemer Venture Partners.

Aghajanyan and Shrivastava see their software as the future of industrial automation, as physical AI often forces a false choice between generalist models and narrow models. They believe their tool is unique because it is general-purpose and flexible, capable of handling various tasks depending on the specific environment it is in.

The algorithmic tool is trained on a million hours of general video, as well as ego and UMI video, which captures scenes from the perspective of a person performing a physical task and records repetitive human actions, respectively. Perceptron thinks their software is well-positioned to lead the wave of industrial automation and is ready to market its software to various industries, including manufacturing, logistics, security, mobility, and media and entertainment.