In San Francisco, ‘autonomous vehicle’ cabs can be hailed via an app as you would an Uber or Zoomy. A visiting Wellingtonian tests it out – and is in no hurry to repeat the experience.
Every couple of years I travel to San Francisco to see my sister, Kelly. While I’m there I always try to make the most of the Bay Area’s reputation as the “home of big tech” and seek out experiences and services that are delivered by robots or otherwise futuristic. Of course, this is all in the name of professional interest – both of our day jobs focus on looking at how technology affects and intersects with people’s lives.
During my most recent visit just last week, I stumbled upon an AI robot manicure service which did a pretty good job of delivering a shiny blue coat to my nails – although it took a call to a human assistant to get the machine going properly and was a real faff to keep my hand still enough for the robot to work. On my previous trip in 2019, Kelly and I headed to Cafe X, a “robot” coffee kiosk that turned out to be a standard push-button coffee machine accompanied by a robot arm that delivered the coffee cup to us with great verve and a funny little dance. Clearly, while the AI robotic future may be arriving in San Francisco, it still needs a fair amount of human assistance and oversight.
The robot experience at the top of my list for my recent visit was to be driven around a busy city in a driverless car. Lucky for me, Kelly had been made it to the top of the waiting list to use the Cruise driverless taxi service but hadn’t tried it yet – she just needed an enthusiastic visitor to get her excited enough to download the app and make a plan to use it.
Driverless cars, or more technically “autonomous vehicles” (AVs), exist on a spectrum from driver-assisted autopilot to cars or trucks that drive unassisted by humans. In Aotearoa, people are starting to dabble with and plan for AVs. For example, local company Ohmio has tested automated shuttles at Christchurch Airport and in other controlled environments and Te Manatū Waka has an automated vehicle work programme. But at the moment, it seems we’re a fair way from having fully autonomous vehicles using public roads, interacting with traffic and pedestrians without user assistance.
In San Francisco, it’s a different story. For the past few years, residents have shared the road with AVs from a number of companies. Until recently, these cars were in testing and training mode and had human driver assistants present in the cars and no passengers, or were driverless but also passengerless. These AVs caused plenty of chaos, with driverless cars frequently spotted stuck in the middle of the road, confusing residents, or even evading police.
Nonetheless, two providers have recently been granted the requisite permits to operate fully autonomous passenger services around San Francisco: Waymo (owned by Google’s parent company Alphabet) and Cruise (a subsidiary of General Motors). Both companies can currently only operate fully autonomous services without a driver present between 10pm and 6am, but only Cruise’s permits allow it to charge for this. However, two draft resolutions from the California Public Utilities Commission which are scheduled to be heard at the end of June would see the now-limited services expanded.
So, last Wednesday night, after dinner in town, Kelly and I walked about 15 minutes into the specified service area, killed an hour at a local bar, and then headed out just after 10pm to catch a Cruise car to as close to the BART (rapid transport) station as possible. The process was pretty easy and will be familiar to anyone who has used Uber or Zoomy. We saw a car was nearby, specified our pick-up and drop-off locations, and within a couple of minutes our car pulled up. So far, so normal.
When the car arrived at the kerb, it was a little unsettling to see no driver inside. I took a bunch of video from our trip and I can be heard excitedly saying “I hate it! I hate it!” as the car pulls up, mostly I assume because it felt uncanny and strange. And perhaps like any technological change or development, the fear of the unknown is more compelling than any actual risk. I mean, how dicey could it be? Kelly unlocked the car (weirdly named “Calamari”) using the app on her phone, and we climbed inside. The app demanded we fasten our seatbelts before departing, and screens embedded in the back of the passenger seats showed the route the car would be taking. We were ready.
Our ride started off well. After pressing the “Start Ride” button on the app, the steering wheel turned to pull out and we were off. A female voice gave us some instructions over the speaker system: keep our seatbelts on, press the “Stop Ride” button on the roof of the car to end our ride early, enjoy our ride. At first it was very weird to see the steering wheel move unassisted, as the car pulled up to four-way stops, paused, and continued when no hazard was sensed. We went up and down hills, gave a wide berth to a pedestrian who was standing on the road, and turned left at a traffic light without too much fuss. I mean, there was fuss, but it was from Kelly and me laughing as hard as we have in ages at an experience that was really unlike anything we’d had before. Every time we spotted a hazard, we asked ourselves if the car would also “see” it and react in time. And it did! It was fine. The feeling I can most equate it to was a rollercoaster, where it’s scary and fun but you know you’re most likely going to be safe.
That feeling changed when, about two-thirds of the way through our ride, we entered a busier part of town close to the central business district. For no reason we could ascertain the car suddenly did a fast swerve towards parked cars before correcting itself. Our mood turned from giddy excitement to a feeling of “oh shit, what did we get ourselves into?”.
As we were closer to downtown there were more cars and people around, meaning more cars and people to act in myriad unpredictable ways. Our car sped up at weird times and did another handful of swerves towards the parked cards on the side of the road. It was legitimately freaky, and I started getting on edge and panicking a bit, telling the car to slow down at least twice and getting stressed at other cars not indicating when turning corners. At one stage Kelly exclaimed “I feel like we’re being held hostage!”. We considered pushing the stop ride button, but stopping on a busy street felt like it might be an even worse idea than continuing.
A few minutes later we arrived at our specified destination. Our car pulled up to the side of the road, told us the ride was complete and we unbuckled our seatbelts and exited. When we were safely on the footpath the car silently pulled away and drove off into the dark city streets ready for its next passengers. We, however, had not finished our journey, and had to walk another 10 minutes to get to the train station due to the limited area in which the cars can operate.
As I write this it’s a couple of days later and I have mixed feelings about our ride. It was genuinely scary at times, and while most of this can be attributed to it being a very new experience, the car did make a series of driving moves that did seem objectively risky. I don’t think I’d jump at the experience again any time soon.
At the moment, the paid AV taxi services in San Francisco aren’t particularly practical for passengers due to the restricted time and area in which they operate. I expect that most users are like me and Kelly: curious folks who want to see what the experience is like. However, autonomous vehicles will no doubt continue to be developed and deployed. Hopefully they’ll get more adept at navigating the unpredictable nature of city streets with variable geography, humans, pets and human-driven cars.
Even as the AV companies are pushing to have their service area and time window expanded, some city politicians and transportation officials in San Francisco are pushing back, asking for more regulation and questioning the safety of these services. It’s true that well-designed driverless cars can reduce some of the risks posed by human drivers: they don’t drive drunk, they don’t text and drive, and they are programmed to follow the road rules (even if they sometimes fail). But they also work best when other road users act in predictable and orderly ways. Which isn’t always the case.
I can’t imagine we will see rides offered to passengers to Pōneke where I live any time soon, except in controlled conditions. Many of the roads are narrow and windy, Aotearoa is largely a “taker” of emerging technology, and regulations and incentives don’t appear to be designed to entice trials here. I may well stand corrected in coming years, and if future AVs are guaranteed to be safer and more efficient than human-driven cars or trucks and can seamlessly coexist with human road users, I won’t complain. But as with all technologies, I don’t think people should just develop them without looking at the bigger picture. In the face of the climate emergency, we need a wider rethink of our transportation system and how we get people and things from A to B. AVs likely have a role to play, but they should only be one part of the picture and not developed and deployed in isolation or at the expense of a system that works for everyone.