Last week, a Nuro self-driving test vehicle parked along the curb across from my house. It reminded me of questions I was asking about autonomous vehicles (AVs) a few years back, when I was mayor of Mountain View. In this case, what would the Nuro robot have done if there had been no vacant curb space?
Mountain View and its neighbors are the center of global autonomous ground vehicle technology. In 2018, when I was mayor, we counted about 20 self-driving vehicle companies in Mountain View alone. As a public office-holder, I had the opportunity to visit and learn about companies such as Nuro, Waymo (part of Alphabet Inc.), and Chinese-owned Didi Labs.
In general, I was impressed by the safe-driving habits of these vehicles, but I felt that our city – as well as communities across t country – was not adequately prepared to address the issues raised by driverless, or in some cases passenger-less, cars and trucks. In the intervening years, I’ve seen little effort to figure out how to integrate such vehicles into the landscape, starting with parking.
If you Google “Kojak parking,” you’ll read that it’s “A parking spot right in front of your destination,” named after the fictional detective, Kojak, who always parked right in front of any building in New York. The same rule applies for “Doris Day parking,” as the actress would always luck out on getting the perfect parking spot in her 1950s and 60s movies.
Like Kojak or Doris Day, the Nuro robot had no trouble pulling into a bare stretch of curb. But what would it have done if other vehicles and objects blocked all other nearby legal parking spots?
If it were controlled by a human driver, like the Amazon Prime and UPS trucks that deliver goods to our residences, it might have blocked a driveway, or even double-parked blocking the narrow street. The driver would have taken a chance that he or she would not be ticketed or towed. He or she might respond to another human driver who asks to pass by or pull out of a driveway.
AVs must be programmed in advance to respond to such situations. One set of rules governs their behavior. Would they circle the block until a space opens up? Would they be programmed to break the law? Would they try to find the spot least likely to upset nearby humans? If asked to move by a resident, or even confronted by a ticket-writing cop, could they even communicate?
This is but one of many challenges that should be resolved before large numbers of autonomous vehicles are deployed on our roads and highways.
Here are just a few of the questions I first raised in 2018:
- Similar to legal, safe parking requirements, will autonomous vehicles be programmed to obey all speed limits, or will they be allowed to mimic human drivers and edge above them?
- What can be done if a vehicle behaves improperly/dangerously? This could be due to poor programming, equipment failure, hacking. The fact that some of these companies do an excellent job doesn’t mean that all do. I can think of one that doesn’t.
- How does a traffic officer ticket a driverless car? How do they exchange insurance information in a scrape or collision? What is “hit and run” for an AV? If there are passengers, are they held accountable?
- Bicycling around Mountain View, I find that many drivers are polite, waving me through an intersection with a smile. Self-driving cars have no similar mechanism for communicating with cyclists or pedestrians. Such indicators would be simple to install and activate, but Waymo told me they were unnecessary. Similarly, will AVs be able to take instructions from people directing traffic, including police?
- Most self-driving cars have multiple cameras. If their cameras “witness” a crime, is the “footage” available to law enforcement? How is their footage kept private? Who decides?
Understandably, government agencies are focused on ensuring that AVs not crash into people, other vehicles, or other objects. Critics – such as people who drive for a living – are focused on the threat of job losses. Both of those challenges are significant.
But if self-driving vehicles are to become ubiquitous, someone will have to figure out how to regulate AVs beyond the state’s safe qualification process. I suggest that cities with AV-test experience convene a dialogue with AV designers and operators, including other interested parties such as insurance companies, to develop model policies. Some of these could eventually be written into law.
Most of these issues that I’ve raised can be resolved, but we can’t rely upon the vehicles to make up the rules, themselves. Once again, technology is driving faster than our ability to govern it.