TOKYO — The Bosch-Daimler-Nivida collaboration announcement Tuesday made clear that it will be a partnership squarely focused on the development of robotaxis.

With plans to launch a test fleet of self-driving taxis in the second half of 2019 in the San Francisco Bay Area, the duo of Bosch and Daimler is set to compete with Waymo, General Motors’ self-driving car unit Cruise, and others.

Among more than two dozen autonomy tech companies, tier ones and OEMs, Waymo has a big head start. Late last year it began testing vehicles in parts of Phoenix without a driver at the wheel — a first for the industry. Cruise has been testing vehicles with safety drivers in San Francisco since 2016.

But really, why the robotaxi rush?

Egil Juliussen, director of research and principal analyst for automotive technology at IHS Markit, describes the robotaxi as “the first autonomous vehicle use case.” Nascent robocars are likely to operate in a geo-fenced area more easily, since in the cab business, “a trip is known before the car is sent for pickup,” he explained.

Fixed routes let AVs drive on city streets with relative safety. The robotaxi business is also easy to scale. AV operators can start business first in a selected area, then gradually spread out.

While running robocars in geo-fenced areas is a topic thus far little discussed, Juliussen believes this approach offers robotaxi operators a chance to accumulate experience with local traffic and behavior. They can “collect real-time data and understand more edge cases,” he added.  

Other benefits include providing consumers first-hand experience with driverless cars. Potentially, it could help alleviate fears of AVs in general, he noted.

Most important, robotaxis make economic sense for AV suppliers, according to Juliussen. By entering the driverless ride-hailing business sooner rather than later, AV vendors can start “bringing in service revenue,” while amassing software and operational data.

Mobility segment cost comparisons

(Source: IHS Markit)

The 64 billion-dollar question, though, is how much revenue can an AV company generate from its never-offered-before robotaxi business.

$6 billion a year?
Juliussen’s quick back-of-the-napkin calculation shows that Waymo could potentially bring in more than $6 billion a year, for example. His estimates assume that Waymo operates 80,000 driverless vehicles every day, each vehicle offering 50 rides per day, a fleetwide total of 4 million daily rides. These numbers assume that Waymo covers three miles a ride at $1.50 per mile.

In this scenario, robotaxis are a nice revenue-generating business for AV vendors who are still working to improve their AV systems.

Juliussen summed up by offering his take on the three best things about robotaxi economics. First, the main savings will be in zero driver costs. Second, most driverless Mobility as a Service (MaaS) vehicles will be electric (Battery Electric Vehicles). They incur “much lower operational costs than internal combustion engine vehicles (ICEV),” Juliussen noted. “Operating cost per mile for BEVs are 50 percent or less of ICEV with gasoline price in the $2.50/gallon range. At higher gasoline prices, the BEV operational cost advantage grows.”

Third, he suspects that driverless MaaS vehicles may also have higher revenue per day per vehicle. Driverless vehicles can offer more hours than a driver. They can even be used to carry freight during slow demand, although this might require a different vehicle type, he added.

However, the real size of this market is hard to guess until robotaxi operators — who aren’t charging their guinea-pig customers today — start asking for money.

Juliussen isn’t alone thinking that the robotaxi is the driving force behind AVs. Phil Magney, founder and principal at VSI Labs, observed, “We see the robotaxi market being targeted by many OEMs since it is pretty certain that by the end of the decade this market will start to take off in urban, geofenced areas.”

Magney, however, feels that the robotaxi market will feature a different kind of car designed for different duty and drive cycles, compared to traditional cars. “Cost will be less of an issue here as the focus will be on safety and redundancies,” he added.

In contrast, Danny Shapiro, Nvidia’s senior director of automotive, noted, “I see robotaxi as only one of the many new autonomy applications that are emerging.”

Of the 370 partners Nvidia is working with today, only 25 are doing robotaxis, Magney said. Besides Daimler and Bosch, Nvidia has partnered with such mobility service companies as Uber, Zoox, nuTonomy and Navya. There will be a different level of autonomy available in different segments of the AV market, Shapiro said. Trucking, currently suffering from a driver shortage, is considered another significant AV market.

When and where to find Robotaxi
The United States is most likely to be the first nation to see Level 4 driverless ride-hailing services emerging. This will be in 2018-2019, IHS Markit said, rapidly moving into a growth period between 2020-2023. Similarly, Level 4 fixed route van services, in campus-like environments, is expected about the same time.

Notably, IHS Markit does not expect Level 4 “personal” AVs to show up until 2022 to 2023. Initially, these will be available only in cities where driverless ride-hailing is permitted. IHS does not see growth in this area until 2024 at the earliest.

As for Level 5 drirverless ride- hailing or personal AVs, IHS is less sure. Juliussen called the launch of such vehicles “uncertain.”

In other words, you won’t be planning a power-point presentation or watching video on the way to work in your own personal AV for at least another decade. Market growth for such sci-fi AVs – either personal or ride-hailing – won’t be happening until 2032-2045.

AV and MaaS deployment is expected to evolve very differently, depending on “current mass transit capabilities, AV regulatory strictness and current success of ride-hailing operators,” Juliussen observed.

By his definition, MaaS includes all mobility services from car-sharing and ride-hailing to ride-sharing/car-pooling and other shared device use (bike-sharing and e-scooter, vans etc.). It is also called transportation-as-a-service (TaaS). MaaS is often used whether the vehicle has a driver or not.

In Japan, for example, L4 fixed-route bus service and fixed ride-hailing will begin in 2020 during the Tokyo Olympics. L4 driverless robotaxis are likely to be the next use-case, starting in 2022.

Curiously, Juliussen sees Level 4 personal AVs appearing in European cities where robotaxis are permitted in 2024.

In China, IHS Markit expects Level 4 driverless ride-hailing will be the first AV use case starting in 2020, while L4 fixed-route van service in campus-like use-cases will also start about the same time.

Evolution of mobility technology
While many in the industry and media initially painted a George Jetson scenario for driverless cars, IHS Markit’s predictions are much more grounded. The market research firm pegs AV mobility in the 2020’s on “geo-fenced, fixed routes, ride-hailing and [a] BEV emerging” market.

The firm does not expect the emergence of AVs that can operate in many routes until 2025 or later, with weather limits for ride hailing and personal use.  

The MaaS Battleground Emerges
By the 2030s, Juliussen believes the MaaS market will unfold around volume driverless MaaS, route-centric driverless MaaS and luxury driverless MaaS. The battleground will be crowded, with fierce fighting among transport network companies, OEMs, mass transit companies and newcomers, including tech companies and startups.

Juliussen assumes that volume driverless MaaS is “where the major service providers will compete,” thus keeping the price per mile low.

Asked why he assumes route-centric driverless MaaS will cost more than volume driverless MaaS, he explained, “I assume route-centric driverless MaaS will have less competition and fewer customers, which would make it more expensive per mile. I also assume the route-centric driverless MaaS vehicles are vans or small buses that cost more, but may not have a full load most of the time. That results in higher prices.”

However, he added, “If these route-specific MaaS have a time-scheduled to get more users per ride, they could be less expensive than volume driverless MaaS. Maybe I need two categories: scheduled and on-demand.”

Juliussen also observed that route-centric driverless MaaS should be distinguished, “because the software complexity is lower and hence many such services are being tested in campus-like environments and they will be an early use-case of driverless MaaS.”

Mobility as a service potential scenarios

(Source: IHS Markit)

Juliussen estimates that by 2023 in the United States, the average price per mile will drop to $0.06, while revenue per mobility as a service AV will grow to be $60,000. Assuming 4.4 million AVs in use by then, nationwide MaaS revenue will total $260 billion, with a 15-30 percent net profit range.

This helps explain why the robotaxi looks — to many investors — like an opportunity too rich to ignore.

— Junko Yoshida, Chief International Correspondent, EE Times