In China and the United States, there is much debate about when and how people will achieve fully autonomous robotaxis on a scale – cars that drive passengers in difficult road conditions without safety drivers driving.
This requires many parts: a huge amount of test data, advanced algorithms, strong operational teams, big checks from investors, support for local politics and many others. Until this day comes, the bold claims of the players on the field seem mostly unattainable.
One recent promise came from impulses one of the most valuable in Asia artificial intelligence startups and the country’s first autonomous driving company achieve unicorn valuation of 1 billion dollars Back in 2018, a four-year-old startup specializing in autonomous vehicle (AV) software solutions recently told TechCrunch that its entire robotax fleet would operate without safety drivers in 2024, while some of its vehicles would already be without drivers . 2022.
The competition in AVs is intense. Waymo Alphabet told customers in October last year, when his cars without a driver are on the road. tesla planned launch the robotaxi network in 2020. In China Toyota backrest Currently, Pony.ai offers autonomous driving services with drivers in two cities. DidB, sponsored by SoftBank, has just begun testing the robotaxi service in Shanghai.
The autonomous taxis that we now see around the world are mainly trial programs operating in designated areas. Most self-managed companies build their own fleet from scratch. Business is bleeding, and commercialization is for many years to come, so the question is who can make it work before the money runs out.
“Expenses [of building car fleets] it’s even unbearable for a multi-billion dollar company like Baidu, not to mention startups like us. But that might be possible for Weimo’s size, ”said founder and CEO Cao Xudong, who appeared in a plain white T-shirt while talking to Zoom.
The 34-year-old founder previously helped set up SenseTime’s face recognition research unit after working at Microsoft’s renowned Asian research unit, which has trained many of China’s leading intellectual brains and entrepreneurs.
Cao said his company could achieve complete vehicle automation while maintaining manageable costs for a startup like himself. While Momenta cannot reveal whether he is actively involved in fundraising, he said that he has a “stable cash flow” that will continue for at least three more years. Company raised over 200 million dollars by 2018.
Before diving into Momenta’s expenses, it’s important to note that not one of its successes can happen without government support. In the process of moving from traditional manufacturing to a technology-based economy, China has provided players in strategic industries such as 5G and artificial intelligence, which, of course, include autonomous driving and large amounts of public funds.
More recently, Beijing moved to accelerate the development of the so-called “new infrastructure” as data centers and 5G networks to offset the economic impact of COVID-19. According to Cao, these are the basic capabilities necessary for the AB, and a political breakthrough will certainly give a powerful impetus to the development of the autonomous driving sector in China.
The government also removes regulatory barriers for promising AV operators. Only this month, Momenta received the first license to attract passengers for its robotaxis on selected public roads in Suzhou, a rich historical city bordering Shanghai with its 4000 square meter headquarters.
A steady path to automation
Unlike many colleagues in their field, Momenta depends on partners who implement technology and collect data, rather than own their own fleet. Although forms of collaboration may vary from case to case, his robotaxi service will be largely a collaborative effort with automakers, who are likely to provide vehicles and, importantly, driver data; local governments that can provide infrastructure such as 5G networks; and himself, who develops self-driving software.
“If you have one million cars, each of which costs several hundred thousand yuan, then this amounts to hundreds of billions of yuan. This is a lot of money, ”Cao said.
Momenta is now working to strengthen his alliance in Suzhou, where we went to one of his trial AVs last year. While a startup aims to achieve full automation, it does not get rid of all security personnel.
“We will take advantage of the 5G infrastructure and we will have remote security personnel, each of whom will control, say, ten cars. Thus, we will reduce the cost of security managers to one tenth of the current level, ”Cao said.
According to the forecasts of the founder of the company, when in 2024 all her cars will be left without a driver, the company will significantly reduce labor costs and achieve a positive operating margin for the car. If everything goes according to plan, he will also transfer his model of light assets to other cities outside Suzhou, entering a period of “tremendous growth.”
“This is a bit like MacDonald’s franchised model. We will develop a set of operational standards and reproduce them in other cities, where we will cooperate with the local government, taxi services, operating companies and so on, ”Cao said.
Momenta also uses less expensive sensors, which the founder called “serial”, such as millimeter-wave radars and high-definition cameras, unlike expensive LiDar sensors. Elon Musk would agree with his choice, saying that “He who relies on lidar is doomed.”
Startup purchases the main components of equipment from international and domestic suppliers, including NXP company, Nvidia and Texas Tools as his semiconductor partners. Cao declined to comment on the consequences of ongoing trade tensions between the US and China, but it is not difficult to understand how the sanctions on the part of D.S. can strangle a startup’s relationship with its suppliers.
Another cost-cutting tactic is automation, which allows the company to minimize the number of engineers. This seemingly simple principle has nuances.
“I have repeatedly told our research and development team that they are not hired as problem solvers, but as architects. What for? Because level 4 [autonomous driving without human input] includes long-tail scenarios, the founder explained enthusiastically. “You may be presented with millions of problems. Of course, we can solve 100 problems with the help of 100 people, but we can’t hire a million engineers to answer a million questions … So if you can create an automatic problem solving system, automation will take care of us. “.
To advance in the AV race, participants must accumulate a large amount of data for training algorithms. Knowing that he lacks the financial ability to deploy thousands of robots, Momenta sells stand-alone driving software to traditional OEM partners and first-tier customers who not only supply him with data, but also provide a steady stream of income.
As soon as partners’ cars enter the market, data about driving a car begin to arrive, and Momenta enters this data into the algorithmic training and periodically updates autonomous cars for consumers.
This setup – getting large amounts of data at low cost – sounds perfect in theory, but it has one big red flag: the data that is the lifeblood of any AI company belongs to car companies, not Momenta. Cao did not seem to be worried, arguing that the partners were interested in data transfer, because Momenta could offer advanced technologies that traditional car manufacturers did not have.
“When we can extract data from customer long-tail problems to train our algorithms, their stand-alone drive systems will therefore be improved. In essence, we create value for customers, ”Cao said with confidence.
Working with outsiders also forces Momenta to combine competing needs. His business is no longer limited to throwing money into research and development. Having customers means that you need to consider what makes commercial sense for car manufacturers, from picking sensors to software solutions.
“The car industry does not think at all like the Internet industry. You cannot ask automakers to adapt to your path, ”says Cao. Thus, he hired a significant number of automotive industry veterans, including business development managers with many years of experience in Mercedes Benz and Toyota.
Momenta was kept on her list of clients, although Cao hinted to us There were few of them last year, because partnerships in the field of AV require close and resource-intensive cooperation.
So far, we know that development of high-definition maps for audio-video cars Toyota, Startup also matters Daimler as a major investorwho began his AV strategy in 2017, although this would not disclose Is the German auto giant a customer? Daimler’s website offers a hint listing Momenta in its portfolio managed Team “M & A Tech Invest”, which is responsible for acquiring technology and startups for the world’s leading brand of premium cars.