A Blog by Jonathan Low


Mar 25, 2017

How Pedestrians May Confound Driverless Vehicles

Humans are just so darned unpredictable. JL

Karinna Hurley reports in Scientific American:

The idea that roads will become safer, with fewer traffic accidents, is a driving force behind the new technology. As pedestrians quickly figure out (driverless) cars’ behavior, they will adapt theirs as well. The effects could be dramatic: instead of more consistent, traffic flow could become chaotic.
Stand at any corner along via del Babuino and it won’t be hard to tell the locals from the tourists. The guidebook holders navigate the swirling vespas and honking Fiats with a mix of hesitation and mad dashing, while the neighborhood residents cross with relative ease; assertive and calm. And it’s not just in Rome: in cities around the world local pedestrians, with a different sense of how drivers will behave, stand apart from occasional visitors. Unless you are on an isolated country road, walking and driving are social interactions, and only residents come to know behavioral customs of their city.
But, a big change might be coming soon. Understanding the psychology of other road users — when and if they will yield — won’t be helpful in cross-walk calculations when the other driver isn’t a person. Self-driving cars are already on the way; by some projections autonomous capability could even be standard by 2030. As drivers, cars will behave differently than humans, and they will almost surely be programmed to avoid hitting people. The idea that roads will become safer, with fewer traffic accidents, is a driving force behind the new technology. But, as pedestrians quickly figure out the cars’ behavior, they will certainly adapt theirs as well. The effects could be dramatic: instead of more consistent, traffic flow could become chaotic.
A recently published paper in the Journal of Planning Education and Research explored how interactions between humans and self-driving cars could change the rules of the road. Author Adam Millard-Ball first explains a current model of how pedestrians decide when to cross the street. Each crossing involves a mental calculation: a choice between crossing as quickly as possible and risking being hit vs. waiting, for who knows how long, or even choosing a new route. Drivers also have a decision to make, to yield or not to yield. The set-up is a cross-walk game of chicken between driver and pedestrian. While intuitively it may seem that pedestrians, more likely to be hurt by a collision, would always yield first, their actions are in fact shaped by social norms. Drivers are likely to yield when hampered by busy traffic or, for example, unpredictable tourists. But, if the local norm is always for pedestrians to wait, the risk of crossing is greater and waiting then makes even more sense.
In a game of cross-walk chicken with a self-driving car, things will be very different. Unlike people, cars will always act predictably; no temptation to glance at a cell phone, need to break-up a fight between squabbling toddlers, or attempts to balance a steering wheel and a drive-through burger. And, cars will almost surely be programed to avoid hitting people. Local customs will be irrelevant; pedestrians will be up against exclusively law-abiding yielders — no matter the corner or block, the pedestrians will have the psychological upper-hand. With full confidence cars will yield, they can be emboldened crossers, even in situations when they do not have the right-of-way. Humans will be free to take advantage of cars.
Millard-Ball outlines three possibilities for how new human-car interactions will alter the roads of the future, starting with pedestrian supremacy. In this scenario, if you need to get somewhere within the center of town you’ll probably go by foot. Your car can drop you off on the outskirts, but will effectively be curbed in urban areas as pedestrians’ impunity to cross streets at their convenience could potentially slow car traffic to a halt. The density of urban areas will continue to increase as walking becomes more efficient than driving.
In the regulatory response outcome, pedestrians will still think twice about crossing the street, but instead of focusing on the risk of being hit by a car, being hit with a potential traffic ticket will come to mind. Attempts to reign in pedestrians could come through a combination of new regulations and infrastructure designed to keep people and cars separated. Planning focused on shared spaces for cars and people will decline and fences and road barriers will increase. Liability for pedestrian-car accidents would primarily fall on the (now) law-breaking pedestrians, and not car manufacturers, further constraining their behavior.
Finally, according to a human driver scenario, the slower travel time incurred by using a self-driving car would outweigh the benefits of a passenger lifestyle. The freedom to check-your email, call-in on a business meeting, or watch Netflix on the drive to work simply won’t be worth taking the extra travel time to get there. Indeed, retaining the advantage in the game of crosswalk chicken will override the convenience of being driven. But, pedestrian-oriented designs only makes sense if most vehicles drive themselves. Ultimately, how neighborhoods evolve to accommodate and incorporate self-driving cars will depend on all the various policy, legal, and technological factors. No matter what scenario prevails, transportation in the future will likely be shaped by the ability of humans to exploit the driverless machines.


Post a Comment