Automated Driving with Intel Mobileye's New Chip

Mobileye to Launch Advanced Supercomputer for Autonomous Vehicles

Intel’s subsidiary, Mobileye, is preparing to introduce a cutting-edge supercomputer to the market. This new system is engineered to equip passenger vehicles, trucks, and SUVs with advanced autonomous driving capabilities.

Introducing the EyeQ Ultra SoC

At the 2022 CES tech trade show, the company unveiled its latest system-on-chip, the EyeQ Ultra. This chip is specifically designed for autonomous driving applications.

The initial silicon for the EyeQ Ultra SoC, boasting a processing capacity of 176 trillion operations per second (TOPS), is anticipated by the end of 2023. Full automotive-grade production is slated for 2025.

Next-Generation EyeQ Chips: EyeQ6L and EyeQ6H

Mobileye also showcased its next-generation EyeQ system-on-chips for advanced driver-assistance systems (ADAS) at CES. These include the EyeQ6L and EyeQ6H.

The EyeQ6L is intended to support Level 2 ADAS features and is expected to begin production in mid-2023.

The EyeQ6H, scheduled for production in 2024, will enable ADAS or partial autonomous vehicle functionality. This more powerful chip will manage all advanced driving assistance functions, multi-camera processing – including parking cameras – and will support third-party applications like parking visualization and driver monitoring.

Mobileye’s History in Automotive Technology

Mobileye is widely recognized as a leading supplier of computer vision technology for ADAS. The company’s first EyeQ chip was launched in 2004, initially focused on collision prevention.

The business has experienced substantial growth, with Mobileye having shipped its 100 millionth EyeQ SoC as of late last year.

A Converging Strategy: Chip Supply and Autonomous Vehicle Development

In recent years, Mobileye has pursued a dual strategy. This involves both supplying automakers with chips for ADAS and independently developing and testing its own autonomous vehicle technology.

In 2018, the company expanded its role beyond a supplier, venturing into the operation of robotaxis.

These two approaches are now merging, aligning with the long-term vision of Mobileye president and CEO Amnon Shashua, who views consumer AVs as the ultimate goal for the industry.

Mobileye’s Full Self-Driving Stack

Mobileye has been actively developing automated vehicle technology for several years. Its comprehensive self-driving stack integrates redundant sensing subsystems utilizing camera, radar, and lidar technologies.

This stack is combined with the REM mapping system and a rules-based Responsibility-Sensitive Safety (RSS) driving policy.

REM Mapping System: Crowdsourced High-Definition Maps

Mobileye’s REM mapping system leverages data collected from consumer and fleet vehicles equipped with its EyeQ4 chip. This data is used to create high-definition maps for both ADAS and autonomous driving systems.

The data collected isn’t video or images, but rather compressed text, amounting to approximately 10 kilobits per kilometer.

This mapping technology, instrumental in the development of the new EyeQ Ultra chip, is accessed via the cloud to provide real-time, up-to-date information on drivable paths.

Data Collection Agreements and Current Coverage

Mobileye has established data collection agreements with six OEMs, including BMW, Nissan, and Volkswagen, utilizing vehicles equipped with the EyeQ4 chip.

Data is also gathered from an aftermarket product sold to commercial fleet operators.

Currently, over 1 million vehicles are contributing REM data, totaling over 25 million kilometers per day, according to Mobileye.

EyeQ Ultra: A Powerful Processing Package

The EyeQ Ultra builds upon previous generations of Mobileye’s SoC architecture. It integrates the processing power equivalent to 10 EyeQ5s into a single package.

Engineered with Mobileye software, the EyeQ Ultra is paired with additional CPU cores, ISPs, and GPUs.

It can process input from two sensing subsystems – a camera-only system and a combined radar and lidar system – as well as the vehicle’s central computing system, the high-definition REM map, and the RSS driving policy software.

Automaker Implementation and Customization

Automakers aiming to offer autonomous driving capabilities in their cars, trucks, and SUVs could utilize this chip to achieve that objective.

The EyeQ Ultra does not include sensors like radar and lidar; instead, it processes incoming information from these sources.

Automakers will have the flexibility to determine how the EyeQ Ultra chip is used, potentially enabling autonomous driving on highways or focusing on automation in urban environments.