On a clear spring morning in Taichung, Taiwan, I rode with a team of researchers and scientists in a 20-seat bus through the campus streets of the world-renowned Industrial Technology Research Institute of Taiwan. As a gold sedan suddenly merged into our lane, the bus came to a jolting stop to allow for a safe right-of-way passage, and I was reminded of the myriad of people who had anticipated this exact moment. Urban planners, designers, sci-fi enthusiasts, journalists, mechanics, and scientists have all awaited the era where a driverless vehicle will successfully react, in real-time, to the seemingly unpredictable high-speed ballet of cars, motorbikes, and pedestrians that all lay claim to the road.
Two days later and more than 11,000km away, a driverless car fatally struck a pedestrian in Phoenix, Arizona. And while the incident did not mark the first death in the world of autonomous vehicles, the pedestrian became the initial casualty in the new frontier of “off-track” driverless cars. When I learned of the event from an innocuous ping on my iPhone while commuting from Hong Kong to Taipei, my response was similar to predictions made by the New York Times less than six months prior. Was this “the one catastrophic accident that could imperil the whole experiment?”
Given the recent headlines, designing trust into autonomous vehicles has become all the more salient. Riders will need to be reassured beyond the statistics, which reveal how human-controlled vehicles are of more danger to the public than driverless cars. But how does trust become automatic, especially in the aftermath of a tragedy? And how can the ownership models pursued by the automotive industry be of benefit?
But how does trust become automatic, especially in the aftermath of a tragedy?
The autonomous, mid-sized bus I rode in Taichung, Taiwan was not considered a Level 5 Automated Vehicle, which is defined as a “full-time performance by an Automated Driving System for all aspects of the dynamic driving task under all roadway and environmental conditions.” In other words, no human driver. The bus I rode in still had a nervous graduate student perched behind the wheel. Anticipating a manual override of any “mistakes” that the artificial intelligence system may have made, human feet were always within inches of the pedals, and other researchers were constantly monitoring the several screens positioned within the body of the bus. It felt like a beta-version of the future, hopefully with less cables.
Though I had previously ridden in a Level 5 autonomous car in Masdar City, Abu Dhabi, the scale of vehicle currently in development in Taiwan (mid-sized buses used for public transportation) re-frames the degrees to which vehicles should be considered autonomous. In fact, accepting vehicles as a type of urban infrastructure positions autonomous public transportation as a shared platform that could provide the same level of accessibility as a private car. Development in Taiwan encourages a shift in scale: no longer should the highest level of autonomous vehicles be designated solely by the sophistication of the artificial intelligence commanding the vehicle. The top benchmark of driverless vehicles should instead be classified by type and not degree of autonomy: vehicles designated for public transit, which have the capacity to reduce the number of private cars on the road and increase urban densification, should be granted the highest accolades. Perhaps a Level 6 should now be designated for completely autonomous vehicles that can transport more than 10 people per trip.
Development in Asia, specifically in Taiwan and Singapore, is focused on driverless public transportation. In the American context, however, technology has focused on simply manufacturing a different version of a single-owner vehicle. But given that autonomous public transportation can leverage discrete infrastructure already present in cities, such as grade-differentiated mass rapid transit lanes, there is a relative ease of implementation compared to private vehicles. Moreover, having autonomous public transportation run in the off-hours of a city could mitigate economic loss for municipal-employed human drivers. Despite this, single-owner autonomous vehicle development in the United States charges forward.
Designing trust into the tech development will come from re-framing the type of autonomous vehicle developed, rather than the degree of autonomy it performs. Implementing autonomous public vehicles, such as mid-size buses, at night would bolster a city’s transportation logistics rather than compete with it. There could also be indirect benefits by offering an alternative means of transportation during peak drinking times in order to prevent drunk driving fatalities. And perhaps the key is to consider that public transportation does not have to be limited to transporting people. Autonomous vehicles can be first implemented for street cleaning, garbage collection, and other tasks performed at off-hours, which would gradually introduce their presence on the road and facilitate regaining public trust. As Marshall Brown, founder of the Driverless City Project forecasts, because “society is cultural, and political, and aesthetic, and about desires — [autonomous developers] are going to need more than just software engineers working on it.”