Recently, I have been in contact with many friends in the industry, in fact, the investment is consistent with the industrial strategy judgment. Taking vehicle manufacturing as an example, before tesla emerged in the field of new energy vehicles, there may be opportunities for new vehicle companies in China in strategic judgment, and then the first round of investment has driven Weilai, Xiaopeng and Ideal, as well as the second echelon of United, Zero Run; and now there is a second wave of investment boom.
On the runway of autonomous driving, the first wave of foreign operations is from Waymo, Uber, Lyft and other operating companies, around the MSSB this kind of travel as a service to do, followed by the United States vehicle companies GM, Ford and the follow-up Toyota, plus the United States IT companies, including Amazon, Microsoft and so on. In this article, I would like to briefly sort out and judge the past, present and future of "investing in autonomous driving".
Figure 1 Overall autonomous driving cooperation map
Part 1: Tech companies with rhythm
From a logical point of view, this wave of investment and research and development in autonomous driving, whether it is Robot Taxi or Trunk, is mainly driven by Waymo, and then Uber and Lyft see the scalability of their business models.
Waymo-driven self-driving taxi commercialization service (Waymo is the boss)
It can be divided into 5 steps——
1. Create prototypes of autonomous vehicles running on specific roads;
2. Obtain an autonomous driving test license and conduct a road test with a safety driver;
3. Launch a small fleet to test passenger operations and gradually expand the geofence;
4. Mass production of autonomous vehicles, upgrade first- and second-generation models, and support large-scale Robot Taxi services;
5. Build and operate the Robot Taxi network and expand to more service areas.
Figure 2 Waymo's test area
At the heart of this is The Robot taxi's strategy of testing and releasing services in restricted geofenced areas, then expanding the area to add more vehicles to the operating fleet. Waymo's first-generation vehicles are based on production versions of conventional cars, enabling autonomous L4 driving through modifications to perception sensors and computing platforms. Waymo began with a focus on first-generation cars, building a factory in Michigan for self-driving car modifications, starting with 60,000 Pacifica Hybrids and 20,000 Jaguar I-Pace EVs.
Note: From the actual situation, this plan can not be completed, especially the I-pace EV is almost out of life.
Waymo then began looking for a second-generation Robot Taxi vehicle — designed for self-driving services (with steering and brakes removed), and of course slower. In this regard, GM's Cruise progress is relatively fast, after all, car companies have their own advantages.
Figure 3 From this pattern it seems too complicated
At present, the biggest problem encountered in the development of autonomous driving is the withdrawal of players, and the failure to see new entrants continue to expand.
Uber was the first to quit, throwing the self-driving service that is constantly burning money and the competition is particularly fierce and can't see when it can make money to Aurora; then Lyft also quit the game and sold the self-driving service to Toyota Woven, as for the technology companies such as Apple, Amazon, and Microsoft that everyone is looking forward to, they still do it around investment in the follow-up, and they did not really participate in the game. In the United States, Waymo and Cruise rushed in front, and there were some followers in the back, while European car companies basically withdrew from the Robot Taxi competition.
In California's DMV report, Waymo became the company with the best total total mileage for self-driving road tests, reaching 2325842 miles, and waymo's fleet expansion increased from 239 vehicles (628839 miles) in 2021 to 693 vehicles in 2021. GM Cruise ran 770049 miles with 137 vehicles in 2020, adding only 1 vehicle in 2021, increasing the mileage to 876104 miles. The third place became Pony.
Figure 4 The autonomous mileage test is almost obsolete
Part 2: Tesla's Approach to Shortening the Path
The Robot Taxi service network released by Tesla in 2019 opened up a new idea for this field by iterating on the path of the models it has already sold. That is to say, there is no need to transform autonomous vehicles so thoroughly, but to adopt shadow mode and obtain more data to iterate around existing models.
Figure 5 Tesla's turning point
The following figure is what I have perfected, and we can see that this competition has returned to the L2 and L3 belts, whether it is possible to run the algorithm through the existing landable functions, and increase the upper limit of the original functions through the iteration of the automatic driving computing power. The change in this logic actually changes the speed (scale) of autonomous driving investment after 2020.
The following parts of the direct change here are -
1) The cost of perception hardware must be controllable: this is around whether it is necessary to do panoramic scanning with expensive mechanical lidar, whether multi-perception redundancy is required, and what price the cost of the entire perception suite is set;
2) Balance of computing power and cloud computing power of the in-car computing platform: Here with the last Tesla AI Day, in fact, a clear idea is given, and the computing power of the car can be coordinated with the cloud server to a certain extent.
3) Software who develops: algorithms can be cut from suppliers, car companies need to have their own integration and optimization departments, along this line of thinking, similar to Volkswagen to give the AID department to Argo and then let their software department cooperate with Bosch, which is a bit of a mechanical division.
Figure 6 Development of software and hardware for autonomous driving kits around production models (incomplete)
brief summary:
I think at the current point in time, around the core of the iteration of the autonomous driving capabilities of passenger cars, the main right returns to the software department and data department of the car company, which actually puts forward requirements for the development of the organizational form in this field.
Figure | network and related screenshots
About author:Zhu Yulong, senior electric vehicle three-electric system and automotive electronics engineer, author of "Automotive Electronics Hardware Design".
Write a message
2030 Mobility Research Laboratory
The first in China to consist entirely of PhDs
In-depth research organization for new mobility in automobiles
·
Zhihu private message ID: Fish is not fish