Last June, a Google deep-learning system that had been shown 10 million images from YouTube videos proved almost twice as good as any previous image recognition effort at identifyingobjects such as cats.
It indicates that most of the processing work - identifying people and other objects in view, and deciding what information to show about them - would likely be carried out by remote computer servers in order to keep the equipment slimline.
The vehicle systems are capable of tasks such as scanning movement of objects around it, identifying a green light from a red light and measuring the trajectory of the vehicle on the road.