The California Department of Motor Vehicles (DMV) recently released its annual Disengagement Report for 2022. As the most valuable reference for self-driving technology worldwide, the report has been losing credibility due to a lack of consistent testing standards. However, its accumulated road testing reports still provide a crucial basis for understanding self-driving technology’s progress and future direction. California's current autonomous vehicle testing permit holders have three levels: the primary level requires a driver to be present in the driver's seat, the intermediate level allows for fully autonomous driving, and the highest level allows for commercial deployment. In 2022, 43 companies have permits to conduct road tests, with 7 achieving the intermediate level: Baidu Apollo, AutoX, Cruise, Nuro, Waymo, WeRide, and Zoox, of which Waymo, Cruise, and Zoox were among the three companies permitted to deploy self-driving technology at the highest level commercially. Of course, these are reference data from road tests in California and do not indicate the three companies have the best self-driving technology in the world; for example, Chinese companies have started withdrawing from California's road tests and have returned to conducting road tests in various cities in China. In addition, many car manufacturers have also withdrawn from the tests. Based on this information, I have summarized a few key observations:
· The top value of MPI (miles per human intervention) has increased significantly from 50,000 miles (AutoX) two years ago to 90,000 miles (Cruise) last year. Although the testing standards differ, overall technological progress is still moving forward.
· Automakers are gradually withdrawing from the autonomous driving race (GM's Cruise still maintains its independence in research and commercialization operations). Currently, consumer-concerned autonomous driving features from major carmakers have only reached Level 3 and are not in the same league as the L4 or L5 full autonomy.
· The top competitors in autonomous driving technology come from different fields: Waymo and Apollo come from the big data and AI support of their parent companies in search engines and are mainly focused on pure technological pursuit (but they also cooperate with multiple car manufacturers); Cruise has deeper connections to the commercialization of vehicle autonomy; Zoox comes from Amazon's logistics application; Nuro focuses on home delivery services; AutoX and WeRide aim to commercialize robot axis.
· The gap in autonomous driving technology development between the two AI superpowers, the United States and China, and the traditional automotive powerhouses of Germany and Japan is growing wider and wider. Of the seven leading teams mentioned above, four are American companies, and the other three are Chinese. The next question is: will these autonomous driving leading teams develop towards cooperation with automobile giants in the future, or will they become increasingly distant and seek different paths?
Although the current trend in self-driving technology applications is towards the logistics industry, where it has lower safety standards (no passengers) or point-to-point public transportation (simpler road conditions) for commercialization to generate revenue to support massive R&D expenses, self-driving technology developers all know that if the ultimate goal of Robotaxi application is not reached, this costly and time-consuming path of self-driving technology development will eventually fail. Because the golden age of L4 and above technology development has lasted for nearly ten years, and the originally planned landing schedule has yet to be fulfilled; as time goes on, the accumulated R&D cost becomes higher, and the difficulty of negotiating cooperation with automakers also increases. Will automakers reign supreme, with self-driving technology relegated to a feature? Or will self-driving technology companies redefine vehicles as "mobile platforms," with self-driving functionality becoming the "soul" of cars and automakers downgrading to low-value hardware suppliers? I have envisioned some possible future trends:
The top teams in autonomous driving technology will continue to burn through large amounts of money in the next three years. Although there will be small-scale commercialization opportunities, it will still be difficult for these companies to break even, leading to a gradual reduction of industry forerunners. Even if they "downgrade" to march on the L2+ market, they still cannot enter the supply chain of car manufacturers. After 2025, dedicated autonomous driving tech development teams will be few and rare. However, in about three years, around 2028, autonomous driving technology will break through the obstacle of "corner cases" and reach a high point of technology for universal application. At that time, if a car manufacturer with similar ideas but not necessarily top sales and profits are willing to seek transformation and form an alliance or joint venture with an autonomous driving company to establish a new car brand, it is highly likely that after 2030, the era of automobiles will be further propelled into a new age of self-driving mobility platforms, rewriting the definition of human travel. This may seem like a distant and abstract idea for most people, but I believe that AI will eventually replace humans as drivers, providing the ultimate answer to the current chaotic traffic and the high number of road accidents. As for the pleasure of driving, people will eventually have to satisfy it through video games.