| Designed from the ground up over the past three years, Tesla's 'full self-drive (FSD)' chipset is based around a neural network that learns how to respond to different road conditions. Rather than using lidar sensors to build up a 3D representation of a vehicle's surroundings, the approach relies on machine vision interpretation and prediction based on inputs from eight high-definition cameras. Already being deployed on current Tesla models in production, the FSD system-on-chip can take advantage of the existing Tesla fleet to improve and refine its autonomous capability, says the firm. Photo: Tesla. |
