To paraphrase Shakespeare, to own or to use, that will be the question in a driverless world. This question was an underlying theme of the latest Smart Driving Car Summit, Who Will Build, Sell and Maintain Driverless Cars? With a collective experience of 150+ years of automotive industry experience, this panel, led by the Dispatcher’s Michael Sena, looked to the past, presented the current situation, and provided informed speculations about the future of driverless.
The Distinction Between Self/Safe Driving and Driverless #
Sena set the stage with a crystal clear definition of driverless and self-driving vehicles. This simple definition collapses the technical complexity of the multiple SAE levels into the following.
- “Driverless cars are those that do not have a human behind the wheel.
- Self-driving cars are highly-automated and can perform most of the driving tasks without human assistance.” Dr. Kornhauser sometimes suggests these are safe driving cars that still require adult supervision.
The scope of this panel was the driverless aspect. As Sena points out, there are two types of driverless vehicles
- Those that fleets will operate and that will have some sort of Operational Design Domain (ODD) restrictions, such as where and when they operate and under what conditions.
- The robot chauffeur, which is the driver for the human who actually owns the vehicle.
Another underlying assumption made by Independent Automotive Researcher Glenn Mercer is that driverless will be electric as well.
Who Will Build Driverless? #
Automotive history may be repeating itself, says Mercer, at least with the number and variety of entities that are vying to be the new kings of the road. In 1910, according to Mercer, there were 250+ OEMs (Original Equipment Manufacturers) in the U.S. Like today, mixed in with the start-ups of yesteryear were established manufacturers from outside the auto industry (bicycles -Peugeot, sewing machines -Opel, and looms -Toyota).
Mercer’s view is that the driverless vehicle has two distinct parts;
- The driverless system, which is really the sensors, compute power, and software that allows it to traverse without a human operator.
- The vehicle, which he believes can be produced by a large number of players.
He sees the high capital costs of the driverless system as leading towards an oligopoly market that can only support a few entities per region. He cites airbags and turbochargers as analogs to today’s driverless systems.
Mercer went on to provide a detailed look at the various manufacturing trade-offs of traditional build options, whether final assembly plants or completely integrated factories. Modular assembly plants, that combine building blocks and micro-factories, are possible options if the economics shift, but Mercer says this is open to debate. Based on comments made by Jay Rogers at last week’s Smart Driving Car Summit, Local Motors would argue that the micro-factory approach combining, 3D printing, building block assembly, and localization could upend the high capital costs of production.
Whether electric or internal combustion, the capital to bring a new car to market is significant and well understood. Mercer cautions that the driverless electric vehicle labor cost may not require significantly less labor than traditional internal combustion engines (ICE). He cites BCG data suggesting a 1% difference in labor hours between ICE and EVs.
As to the question of who will build the physical cars, he sees a variety of players who can build the physical vehicles ranging from:
- Incumbent OEMs (GM, Toyota, etc.) who continue to build vehicles themselves
- New entrants with highly vertically integrated approaches (e.g. Tesla, Lucid)
- Low vertical integration producers that outsource the components and do the final assembly themselves or contract out the entire vehicle to an OEM (e.g. Geely, Foxconn)
But Who Will Maintain and Sell Them? #
“Dealers don’t look very far into the future,” is the well-informed opinion of Sheldon Sandler, CEO of Bel Air Partners. Sandler speaks from decades of experience, including leading the first public offering of a dealership group in the 1990s.
Today, auto dealerships employ over a million people and generate over one trillion dollars in sales. These sales come from five sources of revenue; new and used vehicle sales, parts and service, collision repair, and finance. Surprisingly, 2020 turned out to be one of the most profitable years ever for dealerships.
Sandler believes that car dealerships are here to stay. They already provide fleet sales and maintenance to some extent, implying that dealers will have a role in OEMs’ future plans for driverless rideshare business and associated ecosystems.
Like all incumbent businesses, there are potential challenges on the horizon. To some extent, these challenges are associated with the implementation of new technologies (e.g. electric, driverless, connected, rideshare) that could diminish the role of the dealership and allow for more of a direct connection between the manufacturer (Tesla, as an example) or service provider (e.g. Lyft).
Although there might be less maintenance and repair revenue from vehicles with electric drivetrains, the complexity associated with driverless systems may increase the need for dealership maintenance. Free-lance automotive journalist, Tom Crosby asks whether the increasing complexity due to automation will cause lower reliability.
Crosby has intimate knowledge of vehicle complexity and reliability, as he has been test-driving, on average, a new car a week for the past 25-years. He finds that “higher-end cars seem to have more problems.” The increasing challenge of reliable software is borne out by Stout’s 2019 Automotive Defect & Recall Report, which suggests that approximately 50% of recalls in 2018 were due to either software or electronic defects (page 18).
And when today’s existing ADAS (Advanced Driver Assistance System) vehicles need repair for damages, the costs can be triple that of vehicles without ADAS, according to AAA. From an employment perspective, this will lead to higher-skilled jobs in maintenance and calibration, as mentioned by Audi’s Alison Pascale and implied in this AAA article.
Who Will Buy & Be Responsible for Driverless? #
This complexity of maintenance and high repair costs are sure to continue with the even more complex driverless systems. Those costs coupled with its higher purchase price make Sandler question whether the average consumer will be able to afford driverless vehicles.
Mercer points out that the market is already bifurcated between those who purchase new cars (approximately 17M sold annually) and those who purchase used cars (approximately 40.8 M). The average income of those purchasing new cars is $105k, according to Mercer. The idea that higher-income people will drive demand for personally-owned driverless vehicles is borne out in GM’s prediction, cited by Mercer, that there will be more personally-owned driverless than fleet-owned vehicles by 2030.
An unknown is the potential cost of liability from the failure of systems of personally-owned driverless vehicles. The private airline industry was cited as an example of how product liability became a significant portion of the cost rising from $50 to $100k per airplane produced from 1962 to 1988, respectively. Congress acted in 1994 (PDF) to limit liability to keep private airplane ownership viable.
The potential of liability arising from driverless raises a number of questions, such as
- Will vehicle OEMs have to wrap the vehicle in some sort of mandatory subscription fee to ensure that the vehicles are maintained and the sensors accurately calibrated?
- Will the vehicles automatically return to the dealer (or other entity) in the middle of the night for an upgrade/calibration, similar to the update of a PC’s operating system?
- Who is ultimately responsible for obeying the rules of the road; the owner or the entity that “sold” the vehicle to the owner.
The topic of liability deserves further discussion and will be a topic of a future Smart Driving Car Summit panel.
One conclusion is that the uncertainties around liability will lead to growth in shared driverless. The University of Texas’ Dr. Kara Kockelman suggests that the long-term cost of shared driverless is approximately 50 cents per mile.¹ At that cost, it will decrease demand for personally-owned vehicles, but won’t be a total replacement.
Early research from 2014 by Kockelman and Dr. Dan Fagnant suggests positive benefits in that a fleet approach could greatly reduce the number of cars needed to move a given number of people within a city.
On the other hand, Kockelman’s more recent research indicates that Vehicle Miles Traveled will actually increase 25 to 50%, as driverless transportation increases demand by making long-distance road travel cheaper (PDF) particularly when it comes to the value of travel time (VOTT).
That is, one recovers time by not having to supervise a vehicle and so a longer journey isn’t necessarily the same burden (e.g one can start their workday immediately upon entering a vehicle). One loser in such a scenario could be the regional airlines, as Kockelman estimates a 50% reduction in passenger-travel miles.
Kockelman’s comments and research provide great fodder for future discussion in the ongoing Smart Driving Car Summit series.²
¹ One topic for a future panel is to reconcile Kockelman’s long-term estimate of costs of $0.50 per mile versus the 20 cents per mile predicted by Local Motors’ Jay Rogers at the February 13th, 2021 panel.
² Another area to explore is around standards. Dr. Mike Dobson, retired Chief Cartographer for Rand McNalley expressed concern about the lack of a common language for positional data between manufacturers. Crosby added that additional standardization is needed for standardizing safe driving technologies (e.g. lane departures, etc.).