What Data Do AV/CAM Fleets Actually Collect?

Headline Findings

AV platforms (Waymo, Mobileye) collect rich sensor data at high frequencies and resolution, generating terabytes per vehicle annually. However, there is a significant gap between what fleets collect internally and what cities currently receive. For CAM governance to work, cities need (a) standardized taxonomies of CAM data types, and (b) clear protocols for data extraction that preserve fleet operational security while enabling city oversight. The sensing capability exists; the data governance architecture does not.

Key Technical Specifications

Waymo 6th-Generation Fleet

• Sensor array: 13 cameras (including 17MP imager with 120–180° field of view), 4 LiDAR units, 6 radar units, external audio receivers
• LiDAR specs: ~300 meters range, 360° coverage, 3D object detection (size, distance, classification)
• Camera specs: >500 meters detection range for pedestrians and traffic control devices; thermal stability across automotive conditions
• Data cadence: 10 Hz (10 frames per second) for Open Dataset segments
• Data output: Millions of laser pulses per second (LiDAR), millions of data points per frame (cameras)

Mobileye EyeQ Platforms

• Sensor integration: Camera-primary architecture (8MP–17MP imagers depending on generation), with LiDAR and radar optional/supplementary
• Road Experience Management: Crowdsourced HD map data from millions of EyeQ-equipped vehicles globally, real-time localization feeds
• Data collection: Cloud-Enhanced ADAS leverages “crowdsourced data from millions of Mobileye-equipped vehicles around the globe every day”
• Operational coverage: >80 OEMs globally equip vehicles with Mobileye chips; data collection is ubiquitous but proprietary

Source Data

Waymo

• Meet the 6th-generation Waymo Driver: Hardware overview
• Waymo Open Dataset: Public dataset with 10 Hz sensor data
• Waymo Sensing Overview: Third-party technical analysis

Mobileye

• The Evolution of EyeQ: Chip architecture and sensor integration
• Mobileye Solutions: Road Experience Management and crowdsourced mapping