Perceptron, founded by two former Meta research scientists, this week launched its latest model, Isaac 0.5. The model is designed to help machines perceive, reason, and act in industrial environments, particularly in guiding visual robots in warehouses or on factory floors, and extracting visual intelligence from videos recorded by robots. Isaac 0.5 was released as an open-source model, with its parameters and training materials available for inspection. The company recently raised $21 million in a funding round led by Bessemer Venture Partners.