Recently the Western Automotive Journalists, or WAJ (I had no idea that was a thing), held their annual Silicon Valley Reinvents the Wheel event at the Computer History Museum in Mountain View. Outside in the parking lot was a broad selection of autonomous vehicles. Here are many of them:

One car was Velodyne's Tesla, with the lidar unit behind the in-cabin rear-view mirror. I asked whether the self-driving-computer in the car was using the lidar, they told me that it was not, it was just connected to a laptop on the passenger seat. It was more a demonstration that lidar has reached the stage that it can become almost invisible, rather than a big unsightly spinning thing on the roof. Most of the other cars had spinning lidar on the roof, as you can see in the photos above (and mostly made by Velodyne). Here's Velodyne's Tesla from the outside and the inside, showing how discreet lidar is becoming:

Sven Beiker

The first presentation was a research overview by Sven Beiker of Silicon Valley Mobility. He opened with a couple of pictures that I'm sure you've already seen. On the left is the Easter parade in New York City in 1900. There is just one car and all the other vehicles are horse drawn. On the right is the parade just 13 years later in 1913, with just a single horse and every other vehicle already a car. 

It's always fun to have a quote that shows a lack of awareness of what is about to happen, and Sven had one from the President of the Institute of British Carriage Manufacturers in 1897:

Most people would still prefer for private use the lifelike animated appearance of well-appointed horse traction to any dead mechanism, no matter how smoothly it glided along.

Sven went back to a 1996 paper that looked at the early growth of the automobile industry. Did all the carriage makers really go bankrupt? Which companies won? The paper is titled The fates of de novo and de alio producers in the American automobile industry 1895-1981. You probably don't know what "de novo" and "de alio" mean either. New companies entering the automobile industry, what we would call a startup, are de novo. Companies expanding into automobiles from somewhere else (carriage makers, engine manufacturers, bicycle companies, etc.) are de alio, or in less classical language, diversifiers.

There were five conclusions in the paper:

1. Diversifiers have an advantage over startups because they have the advantage of more resources and more experience.
2. Startups with extensive pre-production stages have a higher chance of success since they enter the market better prepared.
3. Experience, resources, and pre-production stages present an advantage early in the process of entering a new industry but diminish over time.
4. For both diversifiers and startups, the chance for success find optimum and a certain firm density and drop with very low or high density.
5. The origin of the diversifiers matters regarding their chances for success but it remains unclear which specific knowledge provides what advantage.

This paper prompted Sven and colleagues at the Stanford Business School to do a similar study. Not horses to automobiles, but human-driven to autonomous. All business school research must have a framework of a 2x2 matrix and here was theirs:

The two columns in the X direction split the market by what product is being offered to consumers, cars that you buy (or lease), versus miles. The two rows in the Y direction split based on whether a human or a computer is doing the driving. The levels of driving are also on the Y-axis, splitting between levels 2 and 3 as to who counts as the driver.

Existing car companies (OEMs in the jargon) are in the top left with selling cars driven by humans. Uber and Lyft are in the top right, selling miles driven by humans. Regulators call this sort of company a TNC, for "transportation network company". There are no companies today in the bottom half since nobody is actually selling self-driving cars or fully self-driving taxi service. But there are several companies with experimental vehicles in the space. Indeed, there are pictures of some of them at the start of this post. There is a general consensus that the industry will end up in the lower right corner eventually, with autonomous vehicles that you don't own, but summon when you need a ride.

Startups, such as Waymo (owned by Alphabet/Google, so not clear whether startup is quite the right word, but they are not an existing OEM) can attempt to enter the lower right box directly.

Existing OEMs have three approaches to get to the lower right. The disruptive approach is to go straight there. Today a company sells cars for human drivers to drive, tomorrow they sell miles as a service. There are  two other approaches. The evolutionary path is to go from selling the cars of today, and gradually adding more and more automation, transitioning ADAS to take over more and more of the driving. Then sell those vehicles to companies that run fleets of robotaxis, or run their own fleets of robotaxis. The disruptive path is to set up ride-sharing companies and start selling miles, and then gradually remove the drivers from those vehicles as the technology improves.

Sven emphasized that this is work in progress when he showed the picture above. The chances for success are on the x-axis, and dependency of success on the y-axis. The dependency axis needs a little explanation. If self-driving doesn't work out or takes much longer to arrive, then traditional automotive companies and tech companies are fine. They have their existing business. But the TNCs must work since they have no business model otherwise. The startups are in the middle somewhere, depending on how well-funded they are.

It turned out that Sven's great-grandfather had one of the first car dealerships. What ultimately enabled the automobile-based mobility industry was sales, gasoline, repair, and maintenance. Renting vehicles since not everyone wanted to commit to buying. Drivers' education since people didn't know how to operate a vehicle. Horse and buggy to automobile was not just a change of vehicle, it was a whole ecosystem change.

So what ecosystem infrastructure is needed to support the future autonomous mobility industry? He pointed out that "we need to start thinking outside the sandbox". We don't just need autonomous vehicles to operate on Central Expressway or in Pittsburgh. We will need communications, data services, maps and information, signage and markings, testing and education, rules and policies (by the way, Miguel Acosta of the CA DMV spoke later in the day, watch for a future post on what he had to say).

One big area that needs urgent attention is the labels we use for various aspects of autonomy since consumers are not experts. Does "lane-keeping assist" mean that it is assisting or that it is doing it for you? He had a story about confusion 20 years ago when dynamic stability control first arrived. Cars had a DSC button to turn it off, but many customers complained that it seemed to have no effect on the radio. There was no education about it, and the naming was not consistent between manufacturers. We face the same thing with many aspects of autonomous driving. This is one area where the automobile industry can get its act together today. As if to emphasize Sven's point, later in the day Velodyne's Laura Wrisley pointed out that the AAA had identified 20 different features that could all be identified as automatic cruise control (ACC).

Kristin Kolodge

That was the perfect segue for Kristin Koloedge of JD Power to talk about the Mobility Confidence Index. She provocatively titled her presentation If You Build It, They Won't Come. JD Power surveyed 5,000 consumers in the US market for electric vehicles, and another 5,000 different consumers about autonomous vehicles.

First, EVs. They started with people's experience. 4% of consumers have owned an EV, 11% have driven one, 22% have been a passenger in one, and 68% have never been in one. They then looked at people's opinions, focusing on people who own an EV and people who have never been in one.

That experience turns out to make a huge difference. Only 1% of people who have owned an EV say there are no advantages over ICE, but 18% of people who have never been in one though there was no advantage.

That turns out to matter a lot, since the bars are equal and opposite about how likely the participants would be to purchase an EV. They run this survey every quarter, and the numbers didn't move over the last couple of quarters.

We have not moved the needle on their level of interest. It is a wake-up call that we need to do something different.

The second survey was about self-driving vehicles. Kristin emphasized that despite self-driving vehicles being (mostly) electric, they surveyed a different group of consumers. Living in Silicon Valley, it seems that self-driving vehicles are something it is impossible to avoid hearing about almost daily. But 26% of consumers admit to no knowledge at all on the topic. They are more comfortable with goods being transported (24% are uncomfortable about goods) than riding in one themselves (31% are uncomfortable with the idea).

Consumers think that they will be safer. But there is concern about the mixed-mode situation, when only some vehicles are self-driving. "It may be chaos". A full 23% of consumers say there will be "no advantage" with autonomous vehicles.

The top disadvantages have been the same for years: errors, hacking, liability in an accident, pedestrian safety.

There is a low likelihood to purchase/lease, although a good portion of consumers are on the fence at "somewhat likely". But 36% are "not at all likely". This might be because people (me, for example) don't ever expect to personally own an autonomous vehicle. When the technology is ready, it will be robotaxis that people use.

Kristin also picked on lane following. Obviously it is a foundation of self-driving. But consumers are putting things together based on their experience, and current lane-following is not perceived as really helping.

We need to make sure consumers have a positive experience today.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.