There are several business models for AVs, but they've all got drawbacks — some more serious than others...
To develop autonomous vehicles (AVs) capable of handling almost every corner case is cruelly hard. Equally tough and yet less often discussed are the art and science of articulating use cases for AVs.
In short, who needs ’em? Where can AVs possibly be uniquely useful, economically feasible and user-desirable?
In projecting the trajectory of an embryonic AV market, the media, analysts and AV industry insiders shouldn’t duck this fundamental and inconvenient question.
Egil Juliussen, director research, infotainment & ADAS at IHS Automotive, divides AV use cases into four categories: Last-mile delivery, autonomous trucks, fixed route AVs and robo-taxis. Understanding nuances among these different uses is essential, since each imposes on AV promoters varying performance, cost and timeline issues.
We asked Juliussen to walk us through what’s happening in each of these AV market segment. Who are the key players, what’s the current status, what tests and development are happening in different regions in the world?
In parallel, EE Times also talked to Erez Dagan, executive vice president for products and strategy at Mobileye. Intel/Mobileye earlier this month acquired Moovit, developer of a popular app that provides users real-time information such public transportation schedules. Why does Moovit matter?
In our interview, Dagan stressed the need to “weave AVs into public transit.”
Dagan might be onto something. An autonomous vehicle doesn’t operate in a vacuum. Its value must be measured in the context of a society’s overall mobility demands and the circumstances in which people need them. Knowing people’s travel behavior (which Moovit claims to do) could help Intel/Mobileye judge the real need for AVs in real time.
Of all the four application categories defined by Juliussen, “The hardest one is robo-taxi,” he said.
Perhaps ironically, leading AV tech suppliers — including Waymo, Cruise (AVs for GM and Honda), Argo (AVs for Ford and Volkswagen) and Aurora (AV for FCA and Hyundai) — all jumped onto autonomy by gunning for the robo-taxi market. Currently, Waymo One is deployed in Arizona, Aptiv in Las Vegas and GM in San Francisco.
AV testing is even more widespread. More than 65 companies are allowed by the California DMV to test with a driver. Argo.ai is testing in Pittsburgh and Florida.
In the robo-taxi segment, according to Juliussen, the United States has a clear lead in VC funding, AV software development and AV testing. While China is also hot for robo-taxis, “The EU lags in AV testing and robo-taxis,” he noted.
Given the strong VC investment in robo-taxis, this has always been regarded as a market segment with huge potential.
And yet, this is also the segment that will be dogged by one big societal question: What if a robo-taxi kills somebody? Who sues whom? And for how much? And is there a jury on earth that would sympathize with a cabbie-less cab?
Answer: If robo-taxis come with no drivers, any fatalities will leave car OEMs with nobody to blame but themselves. This prospect shatters every existing liability norm.
Carmakers in the past typically dawdled in embracing new safety measures sought by society — such as seat belts — and fiercely opposed regulatory mandates. But with AV? They can’t afford to ignore safety issues, because any failure will point exclusively to them.
Juliussen puts autonomous trucks in a category with “early promise.” The trucking industry tends to agree.
In a recent interview with EE Times, Shawn Kerrigan, COO of Plus.ai, emphasized that trucks are “one of the first places” where AVs will emerge.
Separating autonomous trucks from other AVs is the business case. “We’ve got economics on our side,” Kerrigan explained. Robo-trucks can potentially result in substantial savings for fleet operators, obviously by eliminating human drivers. But that’s just a start. Think about operating cost. Autonomous trucks don’t get tired. They can not only drive many hours without breaks, but also — most likely — would suffer fewer accidents.
Fueling the demand for autonomous trucks is a serious shortage of drivers worldwide.
Technically speaking, “hub-to-hub operations” should be easier to manage, said Juliussen, because “90 percent of the route for autonomous trucks is on highways.” Kerrigan concurred. Since autonomous trucks don’t do much urban driving, they will meet fewer “social problems” in interacting with pedestrians and cars.
Juliussen described autonomous trucks as “only a small market segment.” But to Plus.ai’s Kerrigan, this market is “very large.” He cited the US trucking industry’s annual revenue of more than $600 billion.
The success of any autonomous truck startup will depend on its business model. For example, Plus.ai, which has a full AV software stack for trucks, does not make trucks. It neither owns nor operates a fleet. Its mission is supplying the building blocks of autonomous truck technology.
This model is best illustrated in a joint venture Plus.ai announced last fall with FAW Jiefang, China’s oldest and largest truck manufacturer. The new JV will develop autonomous trucks for China. Its first product, built on Plus.ai’s L4 stack, won’t be an L4 truck, but an L2 semi-autonomous truck.
In other words, despite growing interest in robo-trucks, one prominent JV isn’t yet betting its future on L4 trucks. Fully autonomous trucks are at least four years from now, said Kerrigan. “We need to make sure that all the scenarios [corner cases] can be handled by autonomous trucks,” he said.
If so, how many years can a robo-truck startup survive without any actual trucks out there?
Starsky Robotics lasted five. The AV truck startup based in San Francisco ran out of money after five years and shut down in mid-March.
Juliussen observed that Starsky tried a non-conformist approach to automated long-haul freight delivery. “I kind of liked what they were doing,” He said. Starsky’s trucks, for example, could be remotely operated by humans when exiting highways and driving local streets to terminals — and vice versa. But in the end, Starsky’s ambitions would have required building an infrastructure to support robo-truck operations, which “might have spooked investors,” said Juliussen.
Plus.ai’s COO told EE Times that his company has thus far raised $100 million. Still, there are many more competitors to Plus.ai. Juliussen listed Embark, Ike Robotics and Kodiak Robotics in the United States.
Meanwhile, TuSimple, backed by UPS, announced a plan in March to expand its freight-hauling pilot program with UPS, using retrofitted trucks for Level 4 autonomous driving.
In Europe, Daimler acquired Torc Robotics last fall and this year announced expanded automated truck testing on public routes in the United States.
Now, let’s not forget that leading AV stack suppliers — Waymo, Aurora and others — whose initial target market segment was robo-taxis are also feeling out autonomous trucking. Kerrigan, however, stressed that at Plus.ai, “Our sole focus is on autonomous trucks.”
Fixed route AVs
Autonomous shuttles designed for fixed routes on campuses, airports, hospitals or theme parks are something most consumers have already seen in operation.
To be brutally honest, however, this is AV’s most boring segment. Fixed-route AVs drive slowly and cautiously. They all look like toasters on wheels. They usually feature a limited number of seats — 10 to 15 — carrying more people than a taxi but fewer than a bus or a train.
For those of us accustomed to hailing a ride “right here and right now,” the inconvenience of AV shuttles differs little from the inconvenience of public mass transit in general.
Nonetheless these vehicles hold promise in industrial applications as shuttles in ports or in mining operations, said Juliussen.
The interest in fixed route AVs lingers. But this AV segment attracts the least VC money, noted Juliussen, partly because demand varies wildly by geographical location. It depends on the robustness of public transit in any given city.
In the United States where mass transportation is spotty, interest in robo-shuttles languishes behind robo-taxis, according to Juliussen.
In countries where public transportation systems are well established, the demand for fixed route AVs could be lukewarm. However, in rural areas of Japan, buses no longer operate, because of decreasing populations. But aging populations there have a dire need for rides to clinics, for example. What if there are on-demand AV shuttle services?
Mass transportation systems, which are never profitable, tend to be subsidized by taxpayer money in most places in the world. As a result, operators of AV shuttles would find it hard to compete with public transit. To survive, AV shuttles would need to figure out a way to “weave themselves into the fabric of public transit,” explained Erez Dagan, executive vice president, product & strategy at Mobileye.
Mobileye acquired Moovit, because they believe that knowing people’s movement habits might help AV shuttle operators to inject “intelligence” into fixed-route fleet services. Dagan asked, “What if operators can reroute a shuttle on demand so that they can bring it to where it is needed?”
The key would be coordinating autonomous shuttles with other transit options available in a city. Perhaps public transit systems might seek such intelligence to offer a more holistic system that includes autonomous shuttles.
In a nutshell, cities could solve their first and last mile transportation problems by incorporating “intelligent” AV shuttles, according to Dagan. He suggested this idea’s potential with a reminder that Lyft and Uber today cater to only one percent of all miles traveled by all means of transportation.
According to Juliussen, autonomous shuttles are the strongest AV segment in Europe, because they offer the best synergy with mass transit. Key players in the United States are May Mobility, Local Motors, Drive.ai acquired by Apple. In Europe, EasyMile and Navya are both growing into global leaders. Continental is partnered with EasyMile.
In China, both EasyMile and Navya are independently involved in more than 30 trials, according to Juliussen. In Japan, last fall, Toyota announced that e–Palette will provide automated, loop-line transportation in the Olympic and Paralympic villages for athletes and related staff.
Last-mile goods AVs
Of all four AV categories, last-mile goods AVs is a segment currently features the most rapid development, with strong VC investment, according to Juliussen.
One of the highest profile companies is a Mountain View delivery startup called Nuro. Last year, the company scored nearly $1 billion from SoftBank. Its early retail customers include grocery giant Kroger Co. and Fry’s Food stores in Arizona.
Nuro has also licensed its self-driving technology to a San Francisco-based autonomous trucking company, Ike Robotics Inc.
Earlier this year, Nuro received a permit from NHTSA to operate 5,000 low-speed electric delivery vehicles, called R2, on public roads. R2 will deliver to restaurants, grocery stores, and other service businesses.
Juliussen described R2 “a half-sized car but it’s tall enough to see the traffic. It has no seat for drivers and no steering wheels.”
Delivery AVs come in myriad shapes and sizes. Starship Technologies, founded in Estonia by two Skype co-founders, has gained a foothold at university campuses in Europe and the United States for last-mile delivery of food and groceries.
Not all U.S. states have approved the use of sidewalk delivery robots, however. Typical regulations include stipulations that they can only operate no faster than 6 mph on crosswalks and sidewalks, and must yield to cyclists and pedestrians. A human remote operator is required to monitor the vehicles. States that have approved it include Virginia, Ohio, Arizona, Florida, Idaho, Utah, Wisconsin and Washington.
Other big guns are joining the fray. Amazon is developing a sidewalk AV. Walmart is piloting a grocery delivery program with autonomous vehicle startup Udelv in Surprise, Arizona. In China, a self-driving logistics vehicle startup Neolix raised $28 million in series A+ round earlier this year. Neolix is a strategic partner of Baidu’s Apollo platform.
For investors, one appealing virtue of a last-mile goods AV running on sidewalks is that it costs a lot less than a driverless eighteen-wheeler roaring down the interstate.