The California DMV has recently released annual disengagement reports (how often autonomous vehicles disengaged from autonomous mode during tests) for the year 2022. The impartiality of the reports was questioned last year because the autonomous vehicle manufacturers did not have a standardized road test program. Another reason the disengagement reports received much less attention this year than before is that Argo AI, a well-known autonomous vehicle unicorn, announced the end of its operations a few months ago as a result of the investments that had gone too far and the absence of a timetable for the implementation of its technology. Even so, the reports from the California DMV are still considered one of the most credible benchmarks worldwide for the evaluation of self-driving technology. The California DMV defines three types of autonomous vehicle testing permit holders, from testing with a driver through driverless testing up to deployment (robotaxi operations). Only seven entities have received Autonomous Vehicle Driverless Testing Permits so far. Four of them are US-based, Waymo, Cruise, Zoox, and Nuro, while the other three are China-based, Baidu Apollo, AutoX, and WeRide. Three entities have been authorized to deploy autonomous vehicles, Waymo, Cruise, and Nuro. Since the road tests do not follow uniform standards, the reports can only be used as a reference. The MPI (miles per intervention) indicator is one of the references for self-driving technology, with Cruise, AutoX, and Zoox in the top segment, and Waymo, Cruise, and Zoox are in the top segment in terms of test scale. It is evident that Chinese technology companies are gradually decreasing their deployment of self-driving commercialization in the U.S. and are instead returning to their home country to continue their self-driving road tests. Interestingly, Apple and several German and Japanese auto companies are also listed in the reports, but they are way behind the aforementioned companies in both test scale and MPI. We can see here that self-driving technology is still dominated by the two AI powers, China and the US, and that the picture is likely to remain unchanged in the next five years.
Overall, it will be difficult to fully commercialize Level 4+ fully autonomous driving technology by 2030, whereas Level 2+ is the optimum short-term operational strategy for the technology companies to generate revenue after dimensionality reduction. Such commercial initiatives are widely seen in China and the US. Although many companies continue to challenge fully autonomous driving technology, I'm afraid fewer than 10 will be able to persist for another five or eight years, and they must form strategic alliances or joint ventures with auto giants if they want to ensure that their long-term investments will yield immediate profits from mass production. On the above premise, I think there are a couple of companies worth our continued attention: Waymo, Cruise, Apollo, and AutoX. Of course, there are also some technology companies that have not enabled fully autonomous driving but have the ability to exploit Level 2+ scenarios (such as Horizon Robotics, a Chinese technology company heavily funded by Volkswagen AG). With sufficient groundwork, these companies will still be able to increase R&D efforts in the field of self-driving technology.
In the next few years, it is expected that the annual reports published by the California DMV will gradually decline in significance owing to the bottleneck encountered in the progress of self-driving technology. Meanwhile, car companies will focus on the application of various commercial scenarios and continue to improve the intelligentization level of their own new models, instead of waiting endlessly for the implementation of full self-driving technology.