Home AI Systems & Tools Research & Articles Contact Us
Computer Vision & Sensing

Mobileye EyeQ6 & SuperVision Vision-First Stack

How Mobileye's silicon architecture and camera-centric perception pipeline enable point-to-point hands-free driving with crowdsourced HD mapping.

Mobileye EyeQ visual camera sensing system mounted on windshield

The Vision-First Architectural Philosophy

Mobileye SuperVision uses 11 high-resolution cameras (including 8 megapixel long-range units) to create a 360-degree surrounds visual field. Powered by dual EyeQ6 High SoCs, the system processes optical sensor feeds directly without requiring active spinning LiDAR systems for point-to-point highway and urban navigation.

Mobileye EyeQ6 High Specs

  • Process Node: 7nm FinFET silicon design
  • Camera Channels: Up to 11 8MP surround video streams
  • Crowd Mapping: Road Experience Management (REM) real-time cloud updates
  • Safety Model: Mathematical Responsibility-Sensitive Safety (RSS) formal verification

Road Experience Management (REM)

Through crowdsourced telemetry from millions of production vehicles on the road, Mobileye continuously updates Roadbook HD maps. This low-bandwidth mapping approach supplies centimeter-accurate lane boundaries and localized traffic signal geometry ahead of the vehicle's optical horizon.