A Rough Road for Self-Driving Cars

Should we cut to the chase and just call them self-crashing cars?

Uber got a lot of unwanted attention on April 1, 2018, when one of its self-driving Volvo XC90s on a road test fatally struck a pedestrian in Tempe, AZ.

Elaine Herzberg, 49, was walking her bicycle across Mill Avenue about 10 p.m. when the Volvo, traveling a reported 38 miles per hour, struck her. Herzberg was crossing mid-block, not within a crosswalk. She was still alive when rescue crews arrived, but died later at a hospital.

Aside from the novelty of the autonomously driven car, what really sets this case apart are two major factors — a backup safety driver was on board at the time, and the incident was caught on film and subsequently released on the Internet, where it went viral.

The 22-second video was taken by cameras facing in two directions. One captured 44-year-old driver Rafaela Vasquez before the impact, and the other captured Herzberg crossing the road and getting hit.

Vasquez appeared clearly bored. She is shown yawning, looking off to the left, and then going into panic mode when she looked forward and saw Herzberg in the headlights. By the time she jumped, the impact had occurred, and she seemed to have no time to even think about taking control of the Volvo.

The video shows Herzberg walking out of the shadows in the median, seemingly coming out of nowhere. It appears as if she was in front of the car in a split second, too quickly for the crash to be avoided.

After seeing the video, Tempe Police Chief Sylvia Moir told the San Francisco Chronicle, “It’s very clear it would have been difficult to avoid this collision in any kind of mode (autonomous or human-driven) based on how she came from the shadows right into the roadway.”

However, with all due respect, it didn’t look that way to my unprofessional eye.

Yes, Herzberg came out of the shadows onto the roadway, but the camera lens may be deceiving. The brightness of the headlights may have caused the camera to underexpose the dark areas to the left side, making it difficult to pick her up until she got into the main part of the light beams.

Also, she was clearly walking the bike, and she made it all the way across the width of the car to be struck by the right side of the car. Human eyes are better at adjusting to light intensity differences, and an alert driver may well have recognized her while still in the shadows. An alert driver could have had time to swerve or hit the brakes as Helzberg crossed the path of the Volvo.

Sgt. Ronald Elcock, speaking for the Tempe Police Department, confirmed that the autonomous Volvo made no effort to stop, slow down or turn before striking Helzberg.

And another one

Just five days later, another autonomous fatality went viral on the Internet. Wei Huang, a 38-year-old resident of San Mateo, CA, died when his Tesla Model X crashed horrifically into the concrete divider on Highway 101 in Mountain View.

Crash photos present a brutal image. The Tesla hit the barrier hard, with the damage magnified when it struck a “crash attenuator,” a safety device designed to cushion impacts, which had been damaged in an earlier crash and not repaired. The car then caught fire. The entire front end of the Tesla was taken off the car, with the photos showing the passenger seats eerily staring at daylight. Huang was pulled out of the Tesla and taken to a hospital, where he later died.

Tesla later reported in a blog that the vehicle was on autopilot, with the adaptive cruise-control follow distance set to the minimum. No explanation was offered about what went wrong, but vehicle logs show that the driver was given several visual warnings, as well as one audible hands-on warning, alerting him that he needed to take control of the car. He didn’t.

Looks like proof

Two things seem clear to me.

First, self-driving technology is nowhere near good enough. Second, the idea of a driver being able to take control in an emergency is just wishful thinking. Vasquez is the prime example. Even though she was fully aware that she was participating in a test of unproven technology, it was all too easy for her to become just a passenger, unable to react in time.

In racing, we talk about being “ahead of the car.” We develop the skill of “living in the future,” where we are focused on what we are going to tell the car to do next — not on what it is now doing or just finished doing. That is what lets us anticipate and avoid problems before they actually occur. Once we become mere passengers, we are constantly trying, ineffectively, to catch up with the car.

The big question

All of which brings one very important question to mind: How many more people are we going to kill trying to perfect this technology? That certainly has to be on people’s minds.

Immediately after the Arizona incident, Governor Doug Ducey ordered Uber to stop all testing of autonomous cars in Arizona — they were only allowed to test them under an executive order Ducey had enthusiastically issued about a month earlier.

Immediately after the California incident, a San Francisco motorcycle cop made national news when he stopped an autonomous car and ticketed it (actually the on-board driver) for coming “too close to a pedestrian.”

Is this the future?

Are we all going to be riding in these things in the future? “Legal Files” posed that question to market expert Scot Hall, who is Executive Vice President of Operations for Swapalease.com.

Swapalease.com brokers auto lease assignments for people who want to unload their leased cars. This gives Hall an interesting vantage point on the auto market.

Hall compared autonomous cars to hybrids and electrics, predicting that autonomous cars would behave in the marketplace more like electrics than hybrids.

He noted that hybrids, led by the Toyota Prius, have been very successful in the marketplace. They have strong appeal to people who are concerned about fuel economy. Electrics have the greatest appeal to people who are very concerned about environmental issues. As a result, they have had a very small level of market penetration.

Electrics also have a rather unusual market characteristic. As a vehicle type, they have the highest incidence of leasing as opposed to outright purchase. The explanation is that people just don’t want to own these cars — they want to be able to just walk away from the car when the lease expires.

Buyers have doubts about long-term reliability. They also know that it is a developing technology, and the electric cars that will be available when the lease expires will be more advanced than the one they are driving today. As a result, used electric car prices are quite low.

Hall sees autonomous cars having potentially broader appeal than electrics, as the freedom from having to do the work of driving might appeal to many people. However, the technology is in its infancy, and constant improvement is expected. Thus, leasing may play a major role in their sales, same as electric cars.

Why the rush?

Why are manufacturers moving toward self-driving cars so quickly?

Hall thinks the motivation may be to stay ahead of the curve. He suspects the industry sees this technology as having some level of inevitability, and manufacturers don’t want to be left behind.

After all, they didn’t see the hybrids making such a big impact, and they were upstaged painfully as Toyota grabbed the lion’s share of the new hybrid market.

“It may well be a defensive business strategy, but that isn’t necessarily a bad business strategy,” Hall said.

Hall talks to lots of Swapalease customers about a lot of automotive subjects. That includes autonomous cars. So far, he hasn’t detected much of a demand for them. Nonetheless, he thinks it is possible for that to change dramatically in a short time.

The crystal ball

If we all end up driving self-piloting cars, here are some thoughts to ponder:

  • Who gets sued in an accident? Manufacturer? Software designer/seller? Driver? In pure autonomous mode, there may not even be a driver.
  • Many see autonomous cars as safer than human-driven cars, but that could only be true when all cars are autonomous. How will our society, as a political matter, tell people they can no longer drive the cars they already own? Fold in the economic likelihood that new autonomous cars will be more expensive than existing cars, and the issue becomes a class struggle.
  • How are the autonomous car developers going to monetize their technology? I don’t see Amazon manufacturing cars. But will they form a joint venture with Lexus to market the “Alexus”?
  • If your autonomous car drops you off at work and then drives itself back home, what will we do with all the downtown parking garages?
  • How bad will the traffic jam be when people summon their autonomous cars to come pick them up, but the cars have to wait for owners who get “hung up” on their way to the street?
  • Will it even make sense to own one? Will all cars be owned and operated by Uber and Lyft?
  • If we are going to have to use private tracks to drive our collector cars, how are we going to get them there?

When this all starts making sense to you, try thinking about self-driving semis. ♦

JOHN DRANEAS is an attorney in Oregon. His comments are general in nature and are not intended to substitute for consultation with an attorney. He can be reached through www.draneaslaw.com.

John Draneas

John Draneas - SCM Columnist

John practices law in the Portland, OR, suburb of Lake Oswego, where he focuses on tax and estate planning, business organizations and transactions, and representation of collector-car owners. He is a past president of the Oregon region of the Porsche Club of America and served as the chairman of the PCA’s 2006 parade. His collection includes two Porsches, a Ferrari, an Alfa, a Lotus, a BMW daily driver, a John Deere tractor — and one increasingly famous Jaguar E-type. This month’s “You Write, We Read” on p. 20 is full of SCMer advice on whether Draneas should restore his Jag.

Posted in Legal Files

1 comments

  1. Glad to see someone have an alternate view on this subject. When I get someone telling me that truck drivers are going to get replaced by self driving trucks I point out that the truck, with a load of beer that crossed the country was surrounded by an entire entourage to prevent anyone from cutting in front of the truck. Why? Per Federal regulations a driver has to maintain one second of distance for every 10 feet of length plus another second for each 10 mph over 50 mph plus another second for unusual loads/weather/environment. So the proper following distance for a 70′ truck at 60 mph at night would be 9 seconds. I had one driver respond that to maintain that distance in Los Angeles “he’d have to be driving backwards.” What a truck driver on the freeway usually sees is just the roof of something that just cut in front of the truck. So the truck would slow to re-create the distance and that would cause cars to continue to cut in front of the truck until it would be completely stopped. Then, given the 0 to 60 time of a class 8 truck being in the 1 1/2 minute area there would likely be vehicles cutting in front of it more than before.But I digress. Now Bluetooth connectivity would be useful. Announcing that a vehicle many cars ahead has stopped/slowed or that a vehicle approaching at a perpendicular angle is not likely to stop, that would be good IMHO.

Comments are closed.