In China and the United States, there is a lot of debate about when and how humans will achieve fully autonomous robotics on a large scale – cars driving passengers in complex road conditions without safety drivers driving .
To achieve this, many elements are necessary: gigantic amounts of test data, advanced algorithms, solid operational teams, large investor checks, support for local policies, to name a few. To date, the bold claims of players on the field seem for the most part out of reach.
A recent engagement came from Momenta, one of Asia’s most valuable artificial intelligence startups and the first autonomous driving company in the country to achieve the $ 1 billion unicorn valuation in 2018. The four-year-old startup specializing in solutions software for autonomous vehicles (AV), recently told ClearTipsNews that its entire robotaxi fleet will operate without a security driver in 2024, while some of its vehicles will already be without a driver by 2022.
Competition in AV is intense. Waymo d’Alphabet told its customers last October that its completely driverless cars “were on the way.” Tesla planned to launch a robotaxi network in 2020. In China, Pony.ai, supported by Toyota, now offers an autonomous transport service with drivers in two cities. Didi, the leader in public transport supported by SoftBank, has just started testing a robotaxi service in Shanghai.
An expensive pursuit
The standalone booths that we now see around the world are mainly test programs performed in designated areas. Most stand-alone companies build their own fleet from scratch. The business is hemorrhaging money and marketing is still years to come, so the question is who can run it before running out of money.
“Spending [of building car fleets] is even unbearable for a multi-billion dollar company like Baidu, not to mention startups like us. But that may be possible for the size of Waymo, ”said founder and CEO Cao Xudong, who appeared in a plain white t-shirt during a Zoom call with us.
The 34-year-old founder previously helped launch the research division of facial recognition giant SenseTime after a stint at Microsoft’s renowned Asia Research arm, which has trained many of China’s leading artificial intelligence brains and entrepreneurs.
Uber’s IPO prospectus revealed that its stand-alone unit was burning up to $ 20 million a month. Waymo’s valuation was reduced by 40% last year by Morgan Stanley, citing concerns over cash consumption.
Cao said his company can achieve complete vehicle automation while keeping costs manageable for a startup like it. Although Momenta could not reveal whether this is an active fundraiser, he said he had a “stable cash flow” that would last at least three more years. The company raised more than $ 200 million in 2018.
Before delving into Momenta’s spending, it’s important to note that none of its progress can happen without state support. In its transition from traditional manufacturing to a technology-driven economy, China has made large sums available to players in strategic sectors such as 5G and artificial intelligence, which of course includes autonomous driving. government-funded money.
More recently, Beijing has accelerated the development of so-called “new infrastructure” such as data centers and 5G networks to offset the economic impact of COVID-19. These are basic installations necessary for AV vehicles, Cao said, and the political push will certainly boost China’s autonomous driving sector.
The government is also removing regulatory barriers for promising audiovisual operators. This month, Momenta obtained the first license to recruit passengers for its robotaxis traveling on the chosen public roads in Suzhou, a rich and historic city bordering Shanghai which shelters its vast seat of 4000 square meters.
A sustainable path to automation
Unlike many peers in its field, Momenta depends on partners to deploy the technology and collect data rather than having its own fleets. Although forms of collaboration may vary on a case-by-case basis, its robotaxi service will be largely a joint effort with car manufacturers, who will likely provide vehicles and, most importantly, driver data; local governments, which can provide infrastructure such as 5G networks; and himself, who is developing autonomous driving software.
“If you have a million cars, which each cost a few hundred thousand RMB, that is hundreds of billions of RMB. It’s not a lot of money, ”said Cao.
Right now, Momenta is working to solidify its alliance in Suzhou, where we rode in one of its test AVs last year. Although the startup aims to achieve full automation over time, it does not get rid of all the security personnel.
“We will take advantage of the 5G infrastructure and have remote security personnel who will monitor each, say, ten cars. Thus, we will reduce the cost of security officials to one-tenth of its current level, “said Cao.
When all of its vehicles are driverless in 2024, the company will have significantly reduced labor costs and achieved a positive operating margin per vehicle, the founder announced. If things go as planned, he will also roll out his light asset model in other cities outside of Suzhou, entering a period of “huge growth”.
“It’s a bit like MacDonald’s franchise model. We will come up with a set of operational standards and replicate them in other cities, where we will work with local government, taxi services, operational companies and so on, ”said Cao.
Momenta also uses cheaper sensors, which the founder called “serial products” such as millimeter wave radars and high-definition cameras, as opposed to expensive LiDar sensors. Elon Musk would agree with his choice, having declared that “whoever relies on lidar is condemned”.
The startup sources basic hardware components from international and national suppliers, including NXP, Nvidia and Texas Instruments as its semiconductor partners. Cao declined to comment on the ramifications of trade tensions between the U.S. and China, but it’s not hard to see how the DC sanctions could stifle the startup’s relationships with its suppliers.
The other cost reduction tactic is automation, which allows the company to minimize the number of engineers. There are nuances in this seemingly simple principle, however.
“I have repeatedly told our R&D team that they are hired not as problem solvers but as architects. Why? Because level 4 [autonomous driving without human input] involves long tail scenarios, ”said the founder enthusiastically. “You can be presented with millions of problems. Sure, we can solve 100 problems with 100 people, but we can’t hire a million engineers to answer a million questions… So if you can build an automatic problem solving system, the automation will take care a lot of the work for us. “
To get ahead in the AV race, candidates must accumulate a large amount of data to form algorithms. Knowing that it does not benefit from the financial prowess of deploying thousands of robots, Momenta has sold autonomous driving software to traditional OEM partners and to level 1 customers, who provide it not only with data, but also constant revenue stream.
Once partner vehicles are put on the market, driving data will start to flow in and Momenta will enter this data into algorithmic training and periodically improve autonomous cars for consumers.
This configuration – getting tons of data at low cost – sounds ideal in theory, but it has a big red flag: the data that drives any AI business belongs to the auto companies, not Momenta. Cao did not seem worried, arguing that the partners are incentivized to submit data because Momenta can offer the advanced technology absent from traditional car manufacturers.
“When we can extract data from customers’ long tail problems to train our algorithms, their autonomous driving systems will therefore be improved.” We essentially create value for customers, ”said Cao with confidence.
Working with strangers also forces Momenta to juggle competing needs. Its activity no longer consists only of injecting money into R&D. Having customers means that he must consider what is commercially sound for car manufacturers, from the choice of sensors to software solutions.
“The automotive industry thinks very differently from the Internet industry. You can’t ask the automakers to adapt your way, ”said Cao. As such, he has hired a considerable number of automotive industry veterans, including business development managers with years of experience at Mercedes Benz and Toyota.
Momenta was reluctant about its client list, although Cao hinted to us last year that there weren’t many because audiovisual partnerships require close, resource-hungry collaboration.
So far, we know that Momenta is developing high definition cards for Toyota AVs. The startup also considers Daimler as a major investor, which launched its audiovisual strategy in 2017, although it does not reveal whether the German automobile giant is a client. Daimler’s website offers a clue, listing Momenta in its portfolio managed by the “M&A Tech Invest” team, which is responsible for technology and startup acquisitions for the world’s leading premium car brand.