NVIDIA has manufactured popular graphics processing units for nearly a quarter century. But its been pushing in a variety of new directions, especially into the automotive field. Through multiple partnerships—with TomTom and IBM, in particular—the firm has cruised into the world of self-driving cars using deep learning and artificial intelligence. Add a supercomputer to a dedicated operating system, HD mapping and new computer vision algorithms and what do you get? The BB8 test vehicle that learns by doing rather than being pre-programmed. Information from the vehicle’s sensors and from road imagery create a real-time “occupancy grid” representing the car within its environment. This enables it to learn how to handle varied conditions much as a human would. Its first experiences in a parking lot and on California roads have enabled BB8 to drive in New Jersey, on the other side of the US. It has performed at night and can handle blind curves, unmarked roads and construction zones requiring delicate maneuvers.