In recent years, automakers have increasingly redirected their technology research and development efforts from traditional new energy technology fields to the boundless possibilities of artificial intelligence. Autonomous driving technology has undeniably emerged as the focal point of this transformation. With the relentless advancement of technology, especially in AI, cars are gradually evolving into intelligent mobile spaces.

In this transformation, autonomous driving technology carries the aspirations of many automakers and serves as a pivotal force driving industry change. This year marks a significant qualitative leap for autonomous driving technology across multiple dimensions. Technologically, sensor accuracy has seen remarkable improvements, while algorithm optimization has made vehicle decision-making more precise and efficient. Furthermore, enhanced computing power ensures the stable operation of autonomous driving systems.

▍Technological Breakthroughs and Accelerated Legislation

By 2025, autonomous driving technology is anticipated to achieve substantial breakthroughs in various aspects. In the realm of sensors, new types of LiDAR, cameras, and millimeter-wave radars continue to emerge, offering significantly enhanced performance. For instance, Mobileye's FMCW-based LiDAR boasts no distance blind spots, high resolution, and low emission power, benefiting from Intel's silicon photonics manufacturing expertise to reduce costs. Additionally, multi-sensor fusion technology enables vehicles to perceive complex environments more accurately and comprehensively, effectively tackling various extreme road conditions and weather scenarios.

At the algorithm level, technologies like deep learning and reinforcement learning have undergone continuous optimization, significantly boosting decision-making and planning capabilities. New algorithmic frameworks, such as multi-modal large models like "End-to-End 2.0" VLA (Vision-Language-Action Model), have emerged, reducing redundant data and computational resources, lowering model complexity, and enhancing the overall efficiency and accuracy of autonomous driving from perception to decision-making.

Regarding chip computing power, the most crucial aspect, autonomous driving chips have witnessed leapfrog development, soaring from single-digit to double-digit TOPS to hundreds or even thousands of TOPS. NVIDIA's new-generation automotive chip, Thor, exemplifies this with a single-chip computing power of up to 2000 TOPS. These high-computing-power chips support massive data processing and complex algorithm operations, enabling autonomous driving systems to respond more swiftly and make more precise decisions, thereby laying a robust technological foundation for the commercialization of autonomous driving.

According to McKinsey's forecast, while China's autonomous driving environment is more complex than other parts of the world, the underlying technologies required for deploying autonomous driving are largely the same globally. Technologies encompassing sensors, computing platforms, motion planning and control, and object analysis are likely to continue being dominated by international technology giants. However, technologies more closely tied to local requirements, such as data clouds, maps and location services, and connectivity functions, necessitate comprehensive localized solutions or hybrid solutions integrating local and global technologies.

Regarding hardware costs, with technological maturity and mass production, the prices of core hardware like LiDAR and chips have dropped significantly. Huawei's ADS4.0 platform aims to reduce the cost of core components like LiDAR, promoting the widespread adoption of autonomous driving technology. Similarly, Sony and Honda have collaborated to develop AI autonomous driving technology, aiming to lower sensor costs. The reduction in hardware costs has lowered the overall cost of autonomous vehicles, making them more competitive in the market.

Furthermore, starting in 2024, governments have enacted regulations to facilitate the commercialization of autonomous driving. In 2025, more supportive regulations will be introduced successively to regulate autonomous driving behavior. For instance, the "Regulations on Autonomous Driving Vehicles in Beijing," effective from April 1, 2025, outlines the overall requirements for autonomous driving operations. This regulation expands autonomous driving application scenarios, covering personal passenger car travel, public buses and trams, taxis, road freight transport, ferry shuttles, sanitation cleaning, among others, providing clear guidelines for autonomous vehicles to operate on roads, reducing legal uncertainty, enhancing market confidence, and promoting unified supervision and standardized development of autonomous driving nationwide.

Previously, Wuhan also issued the "Regulations on the Promotion of the Development of Intelligent Connected Vehicles in Wuhan," clarifying guidance on safety responsibility identification for intelligent connected and autonomous vehicles. For example, in the event of an accident, vehicles without drivers or on-board safety officers will be held responsible by the vehicle owner or manager, providing legal safeguards for the local application of autonomous driving.

The improvement of policies and regulations has fostered a conducive policy environment for the commercialization of autonomous driving, accelerating its implementation process.

▍Steady Progress in Commercialization Expansion

In terms of market applications across multiple fields, autonomous driving is continuously expanding its reach. In the urban travel sector, autonomous ride-hailing vehicles have already commenced commercial demonstration operations and tests in cities such as Wuhan, Chongqing, Beijing, Shanghai, and Guangzhou. Wuhan, for instance, as the world's largest unmanned driving operation service area, has not only integrated autonomous ride-hailing vehicles into citizens' daily lives but has also witnessed a steady increase in order volume and service trips.

Recently, GAC AION and Pony.ai forged a strategic cooperation agreement, planning to launch at least a thousand AION Robotaxi models and achieve a positive gross margin for single-vehicle operations. The first batch of a thousand vehicles will be deployed in the Greater Bay Area in 2025, with plans to cover more regions in the future. Thanks to technological advancements and cost reductions, the conditions for the commercial operation of Robotaxi are becoming increasingly favorable, presenting potential for promotion in more cities, thereby transforming people's travel patterns.

As of now, the cumulative number of domestic autonomous driving travel service orders has surpassed 2.5 million, serving over 3.3 million people. Additionally, Beijing has demonstrated eight types of autonomous driving urban application scenarios, including autonomous "taxis" and unmanned delivery vehicles, providing normalized travel and lifestyle services to over ten million people.

The logistics and distribution sector is also a crucial application scenario for autonomous driving technology. In warehousing, unmanned forklifts achieve automated loading and unloading; in trunk line transportation, unmanned trucks handle long-distance cargo transportation tasks; and in urban and last-mile delivery, unmanned delivery vehicles and drones/robots effectively enhance logistics efficiency and reduce labor costs. E-commerce giants like JD.com have already piloted autonomous driving logistics and distribution in certain regions, achieving commendable results.

In port operations, Shanghai Port, Tianjin Port, Shenzhen Mawan Port, among others, also focus on the application of intelligent container trucks and unmanned container transfer vehicles, improving cargo handling and transportation efficiency, reducing human dependency, and elevating the level of intelligence in port operations.

However, despite these applications' accomplishments, they face numerous challenges. At the technical level, the stability and reliability of autonomous driving systems need to be enhanced under complex road conditions and extreme weather. Regarding laws and regulations, global legal norms for autonomous driving are still imperfect, and issues such as responsibility definition and safety standards require urgent clarification. Although current regulations and policies are continuously being refined, imperfections persist. Differences in regulatory standards across regions pose obstacles to the cross-regional operations of autonomous driving. In terms of social acceptance, consumers have concerns about the safety of autonomous driving, and trust takes time to build.

The commercialization prospects of autonomous driving are vast, with application scenarios continually expanding. Industry experts believe that, in addition to travel, logistics, and sanitation, autonomous driving will also play a significant role in medical rescue, public transportation, special operations, and other fields. The core challenge lies in balancing high research and development costs with commercialization.

Experts indicate that the autonomous driving process for commercial vehicles will progress faster than that for passenger vehicles. While it may increase the overall vehicle cost, the savings in labor costs can easily offset this, making it easier to achieve a commercial closed loop and scale up. For many intelligent driving automakers, who will take the lead in achieving commercial breakthroughs remains to be seen. According to the prediction of Horizon Robotics' founder and CEO Yu Kai, high-level autonomous driving will achieve "hands-off" driving in the next three years, "eyes-off" driving in five years, and "minds-off" driving in ten years.

Typeset by Yang Shuo

Image Source: Shutterstock