Text | Wired Insight, author | Wang Huiying, editor | Midnight
Across the ocean, a new story is happening in the field of autonomous driving.
On October 11, Beijing time, Tesla held the long-delayed Robotaxi conference at the Warner Bros. Studio in Los Angeles, USA. With the theme of “We, Robots”, a two-door coupe Cybercab without a steering wheel and pedals was officially launched. .
According to Musk, Cybercab can be charged inductively at US$0.3-0.4 per mile. The cost is expected to be less than US$30,000. It is expected to be officially launched in 2026.
Now, Musk’s dream of a decade ago is one step closer to reality – for Tesla to build a vehicle that can drive fully autonomously without any human intervention.
This can’t help but remind the market of China’s Carrot Run. In 2022, Luobo Kuaipao released its sixth-generation autonomous vehicle, which supports two modes: steering wheelless and steering wheel.
It is reported that the mass production price of Luobo Kuaipao’s sixth-generation unmanned vehicle is less than US$30,000. It has started road testing and is about to be put into use. As of July this year, it has sold 7 million orders across the country. On October 9, it was reported that Baidu Its self-driving travel service platform “Luobo Kuaipao” is currently planning a global layout. It has conducted in-depth communications with a number of international companies and plans to enter overseas markets.
This time, whether it is the form of unmanned vehicles or the implementation of autonomous driving tests, under the comprehensive coverage of China’s vehicle-road-cloud integration and 5G, “Turadish” is at the forefront.
Of course, in addition to running fast, you must also run steadily. Chinese Robotaxi companies are facing the “two heroes” of the United States. Tesla’s Cybercab may be launched in 2026, and self-driving veteran Waymo announced that its driverless taxi service in Austin, Texas, will soon be open to the public.
As a result, the story in the field of autonomous driving becomes more interesting and the competition becomes more intense. The “Tuyang Carrot” racing competition has begun, and the situation of the “three major companies” of Tesla, Carrot Run, and Waymo has also been formed.
In fact, the situation of the “three major companies” is the epitome of the continuous breakthroughs in the autonomous driving industry, and it is also a reflection of human technological progress. As autonomous driving technology continues to make breakthroughs, various companies are looking for technological paths that are more suitable for them and provide their own solutions for humans to free their hands.
More importantly, for a long time, whether it is lidar or chip support, every step of autonomous driving requires real money support. Getting the application implemented and profitable is a difficult problem for players.
Autonomous driving is the general trend, but this road is not easy. Who can run fast and steadily is what the market is paying more attention to.
A battle about speed, technology and commercialization is quietly beginning.
1. The Robotaxi racing competition has started. What are the advantages of “Carrot” and “Carrot”?
Robotaxi companies are meeting each other on a narrow road.
In this competition with Robotaxi as the focus, China and the United States have always stood in the “C position” and are in the leading position. For a long time, as the two countries with the largest number of autonomous driving companies in the world, competition in the autonomous driving field between China and the United States has become increasingly fierce.
When it comes to autonomous driving, Waymo has to mention that this “veteran” player has been driving on the road of autonomous driving for 15 years. It took Waymo seven years to go from a small project within Google to officially becoming an independent subsidiary of Alphabet. In 2016, Waymo conducted its first self-driving test in Phoenix, USA.
With the endorsement of Google technology, Waymo is more like the “academic school” of autonomous driving.
Waymo has always been a representative of the “smart bicycle” route. The so-called intelligent bicycle route mainly enhances the vehicle’s perception, calculation and decision-making capabilities by installing a variety of sensors on the vehicle, such as cameras, lidar and millimeter wave radar, thereby achieving autonomous driving.
Image source Waymo official website
This route enables the vehicle to navigate and make decisions autonomously by simulating the behavior of a human driver, similar to a robot driver with “eyes.” In this way, Waymo is committed to building the most experienced drivers in the world and heading directly towards L4 level autonomous driving.
However, the cost pressure of smart bicycle routes is not small. Not only did it burn through US$5 billion in three years, but its parent company Alphabet continues to inject blood into it. What’s even more troublesome is that Waymo’s data sources all rely on self-driving fleet data and annotated data to obtain feedback. The data source is relatively single. In addition, a safe zone is set up in the geo-fenced test area, and the business expansion speed is relatively limited.
If Waymo is using the hardware system to provide a pair of bright “eyes” for Robotaxi, then Tesla is using software data to create a flexible “brain” for Robotaxi.
Tesla adopts the FSD path, that is, no high-precision map + pure visual solution, and vehicle control is carried out through neural networks. This is an “end-to-end” system, which places higher demands on the computing power of the system compared to the emphasis on technology in bicycle intelligence.
Based on this solution, engineers no longer need to tell the system the specific codes of roads, pedestrians, and traffic lights. Instead, they use large model technology to let the neural network learn how humans drive and find driving patterns through a large amount of driving data and scenarios. For example, if a Tesla hits a dog today, all Teslas will know to avoid the dog tomorrow. In the long run, Robotaxi will become as skilled as an experienced human driver.
Many opinions believe that Tesla has a data scale advantage. As of July 2024, Tesla FSD’s cumulative mileage has exceeded 1.6 billion miles. Under this path, FSD will become smarter and smarter, but the real traffic situation is complex. How to accurately judge and adapt to changes is destined to be a continuous learning process.
Of course, whether it is Waymo’s intelligent bicycle route or Tesla’s FSD solution, behind this is the reflection of American autonomous driving technology’s leading advantage in integrated circuit technology and high-end chip design.
However, although China’s autonomous driving research and development started late, it is more like an “inspirational drama.” A research report from Huajin Securities pointed out that domestic Robotaxi companies are mainly divided into two categories: autonomous driving technology companies and traditional manufacturers. From the perspective of landing cities and test mileage, the progress of technology companies represented by Luobo Kuaipao, WeRide, and Pony.ai More advanced.
In particular, Luobo Kuaipao relies on Baidu’s deep accumulation in the field of artificial intelligence and shows China’s speed to the world with its “bicycle intelligence + vehicle-road collaboration” path.
In short, this path is what we often call “vehicle-road-cloud integration”, that is, on the basis of bicycle intelligence, road test perception is added, and vehicle-end and road-end data complement each other and form synergy, allowing “smart” The car is driving on the road of “wisdom”.
Compared with routes that rely solely on bicycle intelligence, the vehicle-road-cloud integration model makes up for Robotaxi’s deficiencies in perception and communication. Robotaxi is like opening a “God’s perspective” and can better adapt to the complex and ever-changing urban traffic network, thus reducing the incidence of misjudgments and traffic accidents.
This is naturally inseparable from China’s policy support and the rapid development of new infrastructure construction. Currently, our country is vigorously promoting the construction of new infrastructure such as 5G, satellite Internet, data centers, and intelligent transportation, and is promoting 5G LTE-V2X technical standards in road reconstruction to support the evolution of LTE-V2X to 5G-v2X.
Faced with difficulties all the way, in the streets and alleys of many cities in China, the Robotaxi of Carrot Run is like a beautiful scenery, attracting the attention of countless people.
In this feast of autonomous driving, both the East and the West are showing their talents.
American “carrots” have occupied a place in the market by virtue of their first-mover advantage and technological strength. China’s “local radishes” are relying on latecomers to catch up at “China speed”.
The dialogue and collision between the two not only promote the rapid development of autonomous driving technology, but also inject new vitality into the prosperity of the global autonomous driving market.
2. Application implementation is the biggest competition point
Whether a mule or a horse has to be pulled out for a walk, the capabilities of the product are key. Whether Robotaxi can get consumers to pay for it is the key.
The application of Robotaxi is like a battle of speed and passion. Among the imaginable smart travel scenarios, Robotaxi is regarded by the industry as the largest entry point for practical applications, and therefore has become a competition point in the global automobile industry. More importantly, from technology research and development to testing and adjustment, every step of real money will be rewarded.
The difficulty in the industry is that due to the vague business model and complex road traffic environment, its application implementation process has been relatively slow, and most of them remain in the “pilot” stage. But in order to truly implement it, large-scale operations are a hurdle that must be overcome.
As early as 2018, Waymo started exploring application implementation. At that time, Waymo first launched the driverless taxi service Waymo One in Phoenix, Arizona, and launched a paid operation service. Since both the technology and scenarios are in their early stages, users who want to experience driverless taxis must first submit an interest form and then wait for Waymo to randomly select one. In 2020, Waymo became the first company to offer a fully self-driving taxi service to the public.
Although Waymo has established itself as the industry leader, it has also used personal experience to prove the difficulties in implementing autonomous driving applications. The most important difficulty is “safety.”
Unlike other technologies, the error tolerance rate in the field of autonomous driving is extremely low. Once an accident occurs, the consequences will be disastrous. In December 2023, two Waymo vehicles collided with a truck on the highway; in February 2024, a Waymo vehicle collided with a cyclist, causing minor injuries; in June, due to the operation of a Jaguar I-Pace self-driving taxi deployed by Waymo The company collided with a telephone pole during the process, and the company issued a recall notice for 672 Jaguar I-Pace self-driving vehicles.
In order to alleviate market concerns about Waymo’s safety, Waymo issued a statement on June 19 saying that its self-driving cars are still 350% better than humans in avoiding crashes.
Through repeated tests and recalls, Waymo is still a high-quality sample of Robotaxi application implementation, and it is trying to further improve the application implementation process through scene expansion.
On October 2, local time, Waymo announced that its driverless taxi service in Austin, Texas, will soon be open to the public. It is reported that the service will first be available to select users through the Waymo One app and will be moved to the Uber app in early 2025.
On October 10, local time, Tesla, which has been brewing for a long time, finally launched the Robotaxi named Cybercab and the driverless Model Y. Musk said the Cybercab does not have a steering wheel or pedals, and Tesla is expected to launch unsupervised full self-driving capabilities in Texas and California next year.
Source Tesla official Weibo
Obviously, Tesla relies on its strong brand influence and user base to quickly occupy the market through Robotaxi services.
The “foreign radish” in the US market is accelerating the implementation of applications, and the “local radish” in the Chinese market is not to be outdone, and is even running faster.
Take Luobo Kuaipao as an example. As the only Robotaxi company in China that has achieved large-scale and regular testing, it has the endorsement of practical application scenarios and large model reconstruction technology.
According to Baidu’s latest financial report, in the second quarter of 2024, Baidu’s autonomous driving service LuoPao Kuaipao supplied approximately 899,000 autonomous driving orders, a year-on-year increase of 26%, and the total service volume has exceeded 7 million orders. Baidu’s sixth-generation self-driving car RT6 has begun large-scale fully unmanned road testing.
What’s even more rare is that Luobo Kuaipao has always regarded “safety” as the top priority in the development of autonomous driving. As of June 2024, its actual road test and demonstration mileage of autonomous driving has exceeded 100 million kilometers, and there has never been a major casualty. Accidents; each unmanned vehicle and its passengers have purchased insurance with a coverage of RMB 5 million. Data from the past two years show that the actual vehicle accident rate is only 1/14 of that of human drivers.
Behind this, it is due to the “bicycle intelligence + vehicle-road coordination” route of Luobo Kuaipao that adapts to China’s complex road conditions, and it is also inseparable from the exploration of relevant supporting scenarios in China.
Picture source: Carrot Run official website
According to public information, 54 provinces and municipalities across the country have issued normative documents for open roads and demonstration applications, of which 28 cities support commercial exploration of autonomous driving and 24 cities support unmanned testing of autonomous driving. During the 2024 World Artificial Intelligence Conference, super first-tier cities such as Shanghai and Shenzhen also joined the list.
As more and more cities gradually open up autonomous driving pilot projects, China’s autonomous driving application scenarios are becoming more diverse.The market drives scale, and scale expansion is implemented. This is undoubtedly a positive cycle.
Under the positive cycle, Luobo Kuaipao has passed the dual assessment of technological maturity and business model, and has reached the critical stage of evaluating global market competition and opportunities to go overseas.
On October 9, according to Nikkei News, Luobo Kuaipao has been actively developing its global layout recently and has had in-depth communications with a number of international companies and plans to enter overseas markets. Previously, Luobo obtained a license for autonomous driving testing in California, USA. This means that in the near future, Carrot Run will not only be seen on Chinese roads, but also in overseas markets.
It is foreseeable that as autonomous driving technology continues to mature and large-scale implementation accelerates, the potential of the Robotaxi market is gradually being released.This is destined to be a major focus in the autonomous driving competition. Whoever can take the lead in running through and establishing related industrial chains will have the ability to define the track. From the perspective of Sino-US technology competition, we can run as fast as a carrot. The Chinese enterprises represented by the Chinese government must accelerate their efforts in an all-round way.
3. Technology iteration and cost control are key
In recent years, Robotaxi companies have had several common “stumbling blocks”: safety, technology, and cost.Technology can both improve safety and reduce costs. It is crucial to use technology to activate the entire Robotaxi link.
In essence, while Robotaxi liberates human hands, it also makes Robotaxi more like a human being. This is also inseparable from the iteration of autonomous driving technology.
In the United States, developed integrated circuit technology and advanced chip technology are the foundation for Waymo and Tesla’s technology iteration.
In August this year, Waymo launched its sixth-generation autonomous driving system, Waymo Driver. The new iteration of Waymo Driver is equipped with 13 cameras, 6 radars, 4 lidar sensors, and a series of external audio receivers. After the upgrade, Waymo Driver can detect obstacles from a greater distance and has higher computing power and environmental data collection resolution.
In addition, Waymo plans to apply this system to new vehicles designed for driverless driving developed in cooperation with Jikrypton. The two parties are currently advancing the San Francisco test of the sixth-generation Waymo Driver.
On the other side, Tesla’s FSD and Robotaxi form a flywheel effect. On the one hand, the FSD route requires a large amount of data feedback, and Robotaxi’s data can help iterate the FSD algorithm; on the other hand, FSD’s technology accumulation can support Robotaxi’s autonomous driving training and verify a variety of scenarios.
In Tesla’s plan, FSD V13 will be launched in October, and the required mileage will be increased by 6 times; in Q1 of 2025, FSD will be launched after regulatory approval in Europe and China; in terms of computing power, Tesla’s AI computing power scale It is accelerating its expansion and plans to reach a computing power level equivalent to nearly 90,000 H100 chips in 2024 to support autonomous driving training.
In China, with the emergence of large models, Carrot Run has made it possible for Robotaxi to be “more human-like”. At this year’s Apollo Day conference, Baidu released the world’s first large-scale model ADFM that supports L4 autonomous driving. Recently, Apollo Autonomous Driving Open Platform 10.0 will be released soon and will be equipped with Baidu’s latest large-scale autonomous driving model ADFM. This upgrade It will greatly improve the safety, intelligence and ease of use of the autonomous driving open platform.
In addition to technology, from the laboratory to the busy traffic, Robotaxi itself is a very costly business, and how to reduce costs is also a factor that players must consider.
As we all know, the total life cycle operation cost of Robotaxi can be divided into vehicle manufacturing cost, safety operation cost and transportation capacity operation cost. High vehicle manufacturing cost is the core reason for the current increase in unit service cost. Lidar sensing technology and high-computing power chips for autonomous driving are both gold-absorbing beasts.
If software technology still requires time and cost iteration, then reducing costs from the hardware side is the consensus of the industry.
In May this year, the cost reduction effect on the hardware side will be very obvious. Taking Luobo Kuaipao as an example, the sixth-generation unmanned vehicle Yichi 06 is priced at only 204,600 yuan, which is a 60% cost reduction compared to the fifth-generation unmanned vehicle.
Luobo Kuaipao sixth-generation unmanned vehicle, picture source Baidu WeChat official account
On the other hand, Tesla Robotaxi plans to use new car assembly technology and produce it in a way similar to assembling Lego bricks to simplify the assembly process and reduce costs.
Even Waymo, which has always been burning money, has a clear path to reduce costs. In its cooperation with Uber, Waymo is trying to share costs. Waymo is responsible for managing sensor hardware, software and passenger support, and Uber is responsible for keeping the vehicle clean and maintained. As the former’s technology improves, costs such as lidar and sensors are expected to increase due to scale. And reduce costs.
Continuous iteration of technology and careful cost calculation have accelerated the process of Robotaxi’s application.
At Baidu’s 2024 second quarter financial report conference call, Baidu founder Robin Li announced that the Robotaxi business will soon achieve regional balance-of-payments.
This is undoubtedly a milestone in the field of autonomous driving applications, and it also gives the industry great confidence.
With the addition of Tesla, the decade-long Robotaxi competition will enter a new stage. Both Chinese and American giants are accelerating according to the established technical routes and landing application routes. Who can come to the end will depend on time to verify.
In any case, in this battle over technology and application implementation, Chinese autonomous driving companies must have a sense of urgency, cultivate their internal skills, and go all out.
(The header image of this article comes from Waymo’s official website, Tesla Weibo and Luobo Kuaipao autonomous driving Weibo.)
For more exciting content, follow Titanium Media’s WeChat ID (ID: taimeiti), or download Titanium Media App
