Introduction

In the rapidly advancing landscape of autonomous driving technology, numerous players have emerged, some prioritizing raw computing power, others embracing the end-to-end paradigm. Amidst this flurry of innovation, Mobileye stands out as a seasoned veteran, quietly pushing the boundaries of the technology with a rational and steadfast approach.

At this year's CES, Mobileye once again showcased its unique perspective and deep expertise, outlining a rational path for the future of autonomous driving.

(For more insights, please visit: "Israeli Automotive Tech Company Mobileye CEO Discusses Autonomous Driving: The Key to Revolutionizing Transportation")

I. The Luminary of Rationality: Mobileye's Dual Evaluation Criteria

Mobileye, an Israeli automotive technology giant, leverages computer vision, machine learning, data analysis, positioning, and urban road network management technologies to develop advanced driver assistance systems and autonomous driving solutions.

In March 2017, Intel acquired Mobileye for $15.3 billion, paying $63.54 per share in cash. In 2022, Mobileye successfully went public on Nasdaq.

In the realm of autonomous driving, Mobileye acts as a level-headed observer, eschewing blind adherence to trends and instead adhering to its own rigorous standards for measuring technological progress.

Professor Amnon Shashua, the founder of Mobileye, introduced two critical evaluation dimensions: accuracy (safety) and recall (availability). These serve as benchmarks for assessing the quality of autonomous driving technology.

Accuracy is closely tied to MTBF (Mean Time Between Failures), also known as the takeover rate, emphasizing the need for autonomous systems to be exceptionally safe and reassuring.

Recall, on the other hand, focuses on the operational conditions, geographical scalability, and cost-effectiveness of the intelligent driving system, ensuring it is user-friendly, economical, and scalable for mass adoption.

In Mobileye's view, only by satisfying both criteria can autonomous driving truly mature and herald the dawn of driverless cars.

II. Bridging the Gap in End-to-End Autonomous Driving: Mobileye's Approach

While end-to-end autonomous driving may seem like the ultimate aspiration, Mobileye cautions that this path is fraught with challenges.

End-to-end systems leverage large AI models to integrate perception, prediction, planning, and control, enabling data-driven autonomy. However, they grapple with issues such as poor interpretability and the difficulty of achieving 100% accuracy.

Mobileye's solution is to infuse abstract concepts into these systems, such as the RSS (Responsibility-Sensitive Safety) model, a mathematical framework that translates human safe driving principles into mathematical formulas and calculations to review system decisions and ensure safety.

Additionally, Mobileye developed the PGF (Primary-Guardian-Fallback) high-order fusion model, which leverages the collaboration of three subsystems to pinpoint and correct errors, minimizing overall system faults.

This approach equips the autonomous driving system with a "correction mechanism," enabling it to learn autonomously while adhering to strict rules, thereby mitigating the risks associated with an exclusive end-to-end approach.

III. Beyond High TOPS: Mobileye's Quest for Efficiency

In the autonomous driving domain, computing power is often seen as the benchmark for technological advancement. However, Mobileye demonstrates through practical actions that high TOPS (Tera Operations Per Second) is not the be-all and end-all.

Mobileye introduces a new metric, FPS (Frames per Second), which aligns more closely with the actual needs of autonomous driving. FPS directly reflects the speed at which the chip processes images and video frames, serving as a key indicator of the real-time performance of autonomous systems.

In chip design, Mobileye employs a fully heterogeneous computing architecture, ensuring a balanced workload distribution and efficient resource utilization.

The EyeQ series of chips, particularly the latest EyeQ6H and the upcoming EyeQ7H, have emerged as standout products in the autonomous driving industry due to their low power consumption, high performance, and robust scalability.

These chips support L2+ intelligent driving solutions and can be scaled up to L3 and L4 high-level autonomous driving, with efficient power consumption management, providing a solid foundation for the large-scale deployment of autonomous vehicles.

IV. Modular Thinking: Accelerating the Commercialization of Autonomous Driving

In the quest to commercialize autonomous driving technology, Mobileye showcases its profound engineering expertise and modular thinking.

The introduction of imaging radar exemplifies Mobileye's modular approach. Imaging radar boasts precise physical detection capabilities and high resolution, enabling the identification of various obstacles in complex scenarios and providing "true redundancy" in perception for autonomous systems.

More importantly, Mobileye's intelligent driving system operates like a set of "LEGO bricks," allowing for flexible combination based on customer needs.

From ADAS solutions to high-level autonomous driving solutions, rapid functional upgrades and expansions can be achieved simply by adding or removing corresponding modules.

This modular design not only reduces costs but also enhances system scalability and flexibility, providing robust support for the commercialization of autonomous driving technology.

V. From China to the World: Mobileye's Global Footprint

In the Chinese market, Mobileye has demonstrated strong competitiveness and influence.

Collaborations with domestic and international automakers such as BYD, Chery, Great Wall, Volkswagen, and Ford have fueled the growth of Mobileye's ADAS and high-level autonomous driving solutions in China.

Notably, the Zeekr 001 and 009 models equipped with the SuperVision solution have garnered widespread market recognition for their exceptional intelligent driving performance.

Furthermore, Mobileye actively supports the globalization efforts of Chinese automakers, leveraging its extensive customer base, deployment experience, and safety compliance advantages.

In his CES keynote, Professor Amnon Shashua affirmed that by 2027, Mobileye will pave the way for the autonomous driving revolution through its Chauffeur and Drive solutions.

VI. Conclusion: The Rational Path to Autonomous Driving

In this era of surging autonomous driving technology, Mobileye, with its rational and steadfast approach, presents an alternative vision for the future of autonomous vehicles.

It refuses to be shackled by the supremacy of computing power or the end-to-end myth, instead adhering to its own standards for measuring technological progress.

Undeterred by the complexity and challenges of autonomous driving technology, Mobileye actively seeks innovative solutions.

It breaks free from the confines of traditional thinking, daring to innovate and push boundaries.

Mobileye's journey demonstrates that rationalists can indeed shine brightly on the path to autonomous driving.

In summary, as regulations and technology continue to evolve, we have reason to believe that autonomous vehicles will one day navigate public roads safely and efficiently. Mobileye's technological advancements and commercialization efforts herald the gradual opening of the final chapter in the story of autonomous driving.

What do you think, dear reader?