Autonomous Cars: How Rushing Things Could Slow Things Down

The recent Uber fatality of bicyclist Elaine Herzberg, struck by an autonomous car fatality in Tempe (Phoenix, Arizona area) raises serious concerns. Bern Grush, an expert in autonomous vehicles, offers a sobering analysis of the situation.

The Tempe police video of the accident is here (warning: graphic). Grush comments that: “Several viewers of the crash video have suggested there was time for the Uber vehicle to brake and/or sufficient lane space behind Herzberg to avoid a collision.” It looks that way to me. That there was no evasive action or apparently no slowing should raise the most serious of concerns.

Grush also indicates the obvious, that cases like the Herzberg fatality will accelerate negative publicity about autonomous cars. This is not at all surprising, given the aggressive implementation narrative that has been adopted by so many. For example, a Stanford University study (according to one report) suggests that car dealerships will be a thing of the past before 2025 and 95 percent of cars will be autonomous by 2030.

There is a (not perfect) parallel. San Francisco’s BART, the Bay Area Rapid Transit system, was to have operated driverless as well. But then, a train ran off the end of the line at Fremont and landed in a parking lot shortly after service began (the “Fremont Flyer”). Autonomous operation of the BART system has never resumed. There have been substantial advances in automated rapid transit. The first systems were within airports. Eventually, Lille, France opened an automated rapid transit system. Now, even the busiest Metro route in Paris (Line 1, La Defense to Vincennes) is fully automated. But there is a big difference between autonomous cars and automated transit systems. The transit systems are designed (or redesigned) from the “ground up” for driverless operation. Autonomous vehicles will not have the luxury of such a controlled environment in the foreseeable future.

At the same time, the public seems to be increasingly concerned about both the operation and conduct of the broader information technology industry. This has been fueled by cases like the Equifax data breach, to the Facebook (and other) privacy concerns and the hacking of international intelligence systems. Many of us have had the unhappy experience of not-ready-for-prime-time PC operating systems, so flawed that they were quickly replaced by entirely new systems. In short, despite the transformative effects of automated technology, premature implementation is more likely to lead to delay than sustainable implementation.

Autonomous cars will doubtless replace self-driving cars. However, people will be enticed, not forced into a driverless future --- when the technologies are ready. The tragic death of Ms. Herzberg suggests that this is later rather than sooner.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Don't make perfect the enemy of the good

There were 110 other people killed by human-driven cars on the same day Ms. Herzberg was killed by the AV. Zero AV fatalities shouldn't be the metric. Rather, AVs should be, say, an order of magnitude safer than human driven cars. 40,000 people in the US were killed in traffic accidents last year. What if AVs can bring that down to 4000/year? Wouldn't that be a very good thing?