Share the Road

Self-driving cars may make intersections less dangerous and more efficient. By communicating with an “intersection master,” AVs might be able to navigate intersections without stopping or hitting other cars. This video from the University of Texas shows how that could work:

But there are some essential aspects of intersections that these designers have utterly ignored – other street users. This intersection design does not seem possible if there are non-AV road users like bikes.

Uber is already having issues with bikes in San Francisco. When making right turns, the Uber AVs “posed a direct threat to cyclists” by illegally and unsafely moving into bike lanes. One proposal for making AV-bike interactions safer is to require bikes to be outfitted with sensors just like AVs are, and then having the bike communicate with the AV. But if AVs can’t navigate city streets safely on their own, the solution is not to force other road users to make up for what the AV lacks.

AVs are not all bad for bikes, however. The extra cautions AVs take compared to human-driven cars reduce the risk of car-bike crashes. AVs aren’t good at the nuances of bike-car interactions, like when bicyclists stop at stop signs but maintain the body language of a moving biker. But since the default AV reaction to uncertainty is caution, the inability to interpret nuances is more annoying than dangerous.

AVs can also reduce the risk the risk of car doors opening into bicycle paths, causing the bikes to crash (a phenomenon known as dooring). AVs could reduce that risk by preventing passengers from leaving the car until the bike lane is clear.

Dedicated bike infrastructure – think cycle tracks, not just bike lanes – could reduce the number of bike-car interactions, but until that happens AVs will need to learn to interact with bikes. This is made more difficult because bikes don’t follow the same laws as cars. In all states but Idaho, bikes are treated like cars and are required to stop at stop signs and red lights. Bikers break those laws so often, however, that when they start to follow them drivers get frustrated and confused.

The relationship between cars and bikes on roads is a barrier to mass-adoption of AVs, but it’s a barrier that has a relatively simple solution (although not necessarily one car-addicted Americans will like). Shifting more road space from cars to bikes could reduce the number of car-bike interactions and make trips more pleasant for everyone. Vancouver has started to accept bikes as an important form of transit, and has made biking a much safer experience.


Sources

https://www.theguardian.com/technology/2016/dec/19/uber-self-driving-cars-bike-lanes-safety-san-francisco

https://archives.sfweekly.com/thesnitch/2015/07/30/this-is-what-happened-when-bicyclists-obeyed-traffic-laws-along-the-wiggle-yesterday

https://sf.streetsblog.org/2012/07/20/bikes-are-not-cars-why-california-needs-an-idaho-stop-law/

http://www.npr.org/sections/alltechconsidered/2017/07/24/537746346/bikes-may-have-to-talk-to-self-driving-cars-for-safetys-sake

https://www.washingtonpost.com/news/innovations/wp/2015/08/26/how-fixed-gear-bikes-can-confuse-googles-self-driving-cars/?utm_term=.fcf2cbb80b9b

No Speed Limits?

Drivers want to go as fast as they please, but speed limits (theoretically) restrict that ability. People regularly ignore the speed limit, instead going as fast as they believe they can without crashing. Often, they are not at fault when they succumb to the desire to speed – traffic engineers often design roads to be safe at much higher speeds than are posted.

Some argue that AVs will be better at sensing and following the design speed, making the posted speed less important. AVs have trouble reading ordinary street signs, so having them rely on the design speed might be a good thing. Design speeds are supposed to be equal to or higher than posted speeds, so allowing AVs to rely on those might increase highways speeds, but not by much as the AASHTO book recommends design speeds around 70 mph, rather than the 65 mph that are often posted.

Signs that confused AVs; the middle sign looked like a speed limit

AV manufacturers are already starting to design against the speed limit. Tesla argues that allowing AVs to drive faster than the speed limit to keep up with traffic is safer for everyone. Google’s Eric Schmit agrees, saying that the “biggest problem” about AVs is that they follow the speed limit. Brad Templeton makes a somewhat detailed argument for why he believes AV operators should be able to set the car to any speed, not limited by the one posted.

I’m open to the idea of removing speed limits on freeways, but many of these proponents don’t make a distinction between speed limits on the highway and on multi-modal streets. If speed limits are going to be removed on highways, I think a fair deal would be to reduce speed limits on multi-modal streets. Having lower speed limits (think 10-15mph) on city streets would allow all users (pedestrians, bikes, buses, cars) to enjoy the space. Having no (or higher) speed limits on highways would also allow all users (cars, sometimes buses) to enjoy the space. When arguing against a speed limit for AVs, the details are essential. It is one thing to have no speed limit when just cars are around, but very different to have no speed limit when cars are sharing the space with other users.


Sources:

https://www.theguardian.com/technology/2017/jan/16/tesla-allows-self-driving-cars-to-break-speed-limit-again

http://www.slate.com/blogs/future_tense/2012/07/13/eric_schmidt_on_self_driving_cars_biggest_problem_they_obey_speed_limits.html

http://www.slate.com/articles/technology/future_tense/2013/10/robocars_do_they_need_speed_limits.html

https://globalnews.ca/news/3654164/altered-stop-signs-fool-self-driving_cars/

https://safety.fhwa.dot.gov/geometric/pubs/mitigationstrategies/chapter3/3_designspeed.cfm

Toddler take the wheel

Self-driving mini-bus line in Tallinn, Estonia.

As with most forms of public transit, Europe is far outpacing the United States. Wageningen, Denmark; Helsinki, Finland; and Tallinn, Estonia have all pushed ahead to see what a future of autonomous public transit might look like.

In January 2016, AV mini-buses in the agricultural town of Wageningen drove “back and forth along the side of a lake” at 5mph. They were, however, the first AVs to “operate without a driver, on a public road.” The WePod mini-buses are expected to eventually reach a maximum speed of 15mph.

In late 2016, Helsinki put two AV mini-buses on the road “alongside traffic and commuters.” Like in Wageningen, the buses travel 7mph, although they have the ability to go up to 25mph. Finland is unique in not requiring “vehicles on public roads to have a driver,” meaning the buses could be filled to capacity (of 12) with passengers who had never experienced an AV.

During July and August 2017, two AV mini-buses served “the center of the town” in Tallinn. They went a maximum of 12mph, and just bounced back and forth between two stops. Each bus costed around $100,000.

I think Tallinn is the most interesting example because it shows more clearly the challenges that AVs face on roads. Much of the discussion surrounding the interactions between AVs and people focuses on the people using the car; we have not really discussed how AVs will interact with people who are not sitting in the car. Within the first three days of AVs starting service in Tallinn, there had been no “major incidents” but quite a few “near misses.” There is only one intersection where the buses interact with other traffic, but at that intersection the buses have broken multiple laws.

They have ignored a pedestrian walk sign, a “speeding police car’s emergency lights,” and a red-light at the intersection. These are all common occurrences that the AVs should be equipped to respond to, and yet they cannot. If AVs ignore pedestrian walk signals, how can we be sure they will stop for pedestrians who crossing the street in crosswalks? How can we be sure they will stop for pedestrians crossing the street unexpectedly (i.e., jaywalking)? AVs will not be widely accepted until they can appropriately react to unexpected but common occurrences like people crossing the street outside of crosswalks.

The good news for AVs is that due to their slow speeds, they are not nearly as dangerous to pedestrians as human driven vehicles. As long as AVs continue to operate below 20mph, the likelihood of them killing a pedestrian is only 5% and there’s a 20% chance that the person they hit would walk away with no major injuries. Human driven vehicles might be more aware of pedestrians, but they are also much more likely to drive at speeds higher than 20mph, no matter what the speed limit is. (One reason for this is the design speed is often slightly higher than the posted speed, making it easy for drivers to speed without noticing.) Getting hit by a car is not inherently dangerous (e.g., a car going 5mph isn’t going to do much damage) – it is only dangerous when the car is driving fast enough to cause damage.

Some think that the need for AVs to always stop when a pedestrian walks in front of them will limit the consumer appeal of AVs. A professor at UC Santa Cruz thinks that the need for AVs to be “risk-averse” will ultimately lead to pedestrians to “act with impunity” and dominate the roads because they will be “secure in the knowledge that a car will yield” once they step into its path. The professor, Adam Millard-Ball, is quoted as saying “From the point of view of a passenger in an automated car, it would be like driving down a street filled with unaccompanied five-year-old children.”

In conclusion: we think driving in a world of AVs will be amazing, but it might actually feel like a 5 year old has taken the wheel.


Many of the examples of self-driving mini-buses seem to have a few things in common:

  • They have an exclusive right-of-way. They might be driving on public roads, but they aren’t necessarily dealing with other cars or pedestrians. Right now it seems like Helsinki is the exception to this.
  • They don’t go very fast. Tallinn’s buses can go 30-40mph, but they operate at 12mph. Some mini-buses go only 5mph.
  • They are not yet fully relied on. Even in Helsinki where the law does not require a driver, the buses have operators to ensure they don’t hit pedestrians.

Helsinki’s buses seem to be the most advanced right now: https://www.youtube.com/watch?v=DFHXFF4BhWw

Paris, where the AVs have an exclusive right-of-way (for now): https://www.youtube.com/watch?v=dfBp-oGNoqE

Some other countries where self-driving minibuses are being tested: Japan, Greece, and Australia.


Links:

http://www.dailymail.co.uk/sciencetech/article-4774580/Driverless-buses-roads-Estonia-s-capital.html

https://www.eu2017.ee/photos/self-driving-bus-line-opening-tallinn

https://e-estonia.com/self-driving-minibuses-set-to-roam-the-streets-of-tallinn-in-2017/

http://www.bbc.com/news/blogs-news-from-elsewhere-40800867

https://www.theguardian.com/technology/2016/aug/18/self-driving-buses-helsinki

https://www.theguardian.com/technology/2016/jan/28/driverless-bus-trial-in-netherlands-will-be-first-on-public-roads

https://news.ucsc.edu/2016/10/pedestrians-self-driving-cars.html

Self-Driving Sprawl?

A recent New York Times article looks at the daily life of an extreme commuter outside San Francisco. Ms. Sheila James makes a two and a half hour commute from Stockton to San Francisco each day – she relies on her car, two trains, and a bus to travel the 80 miles from home to work. Other commuters eschew public transit and drive this route instead. Alert at the wheel before the sun has risen, their commute is in some ways just as hard as Ms. James’s, even if it might not be quite as long.

The introduction of AVs could significantly improve the lives of super-commuters like Ms. James. Instead of leaving the house before 4am to take transit to work, she could schedule an AV chauffeur to pick her up and drive the 80 miles while she slept – or at least relaxed, because even if the vehicle was L3 rather than L4 she wouldn’t have to actively drive the entire way. If other commuters near her Stockton home were also heading to San Francisco (as they likely were), Ms. James and her neighbors could commute together in a large (L4) AV and sleep, read, or otherwise relax.

There is, however, a downside to making super-commutes like Ms. James’s easier.

Exurbs, the suburbs of suburbs, are growing.

If AVs make lengthy commutes easier, workers may choose to live further and further from their jobs. Right now people are limited in how far from work they can live based on their willingness and ability to commute long distances. If AVs make long commutes less of a nuisance, it is possible that people will continue to move further away from cities into increasingly large and sprawling exurbs. (Exurbs are most easily described as the suburbs of suburbs, and they’re growing quickly.)

Houston Post-Harvey, showing the risk of sprawl building on wetlands.

As cities become more and more sprawling, they often run into land that cannot be wisely and safety built on. The disastrous results of unregulated sprawl was seen most recently when Hurricane Harvey hit Houston, destroying entire neighborhoods that had be built in questionable areas like in-filled wetlands. Houston’s sprawl and lack of land use planning was criticized as being one reason the hurricane was so disastrous.

Urban planners and transportation experts are already thinking of policy measures to reduce the likelihood that AVs help sprawl spread. At a recent 3 Revolutions Conference hosted by the Institute of Transportation Studies UC Davis, some presenters suggested “pricing mobility” as a way to limit sprawl. “Pricing mobility” would involve taxing “VMT [vehicle miles traveled], vehicle size and incentivizing increased vehicle occupancy” (pg. 3) This pricing scheme could work whether cars were personally owned or temporarily rented. Assuming AVs were treated like a ride-sharing service, riders could pay for the VMT as part of their rental fee, receive discounts if they ride with others (like what happens with UberPool), and have a higher base rate if they request a larger vehicle. Assuming AVs were private vehicles, these policies would simply raise the percentage of total costs imposed on society that vehicle owners would pay – it would not necessarily raise the total cost of cars (whether AV or human-driven) but would simply shift more of the total cost burden onto the person benefiting from the vehicle, rather than continuing to have society grant drivers enormous subsidies.

Differing costs to society; the exact numbers might change with AVs (like a potential reduction in accident risk) but cars are still the most expensive form of transportation.

AVs will reduce the human labor costs of driving, with outcomes that are both positive and negative. On one hand, it could improve the lives of individual people like Ms. James who commute 80 miles out of necessity because they cannot afford to live closer to their jobs. On the other hand, the potential societal costs of AVs include more and more sprawl, which can hurt society economically and environmentally. (Even if AVs are electric and so might be less carbon intensive, very low density development like urban sprawl would still hurt the environment by consuming land and resources that might otherwise be left alone.)

Quick Intro

I’m a junior majoring in Urban Planning. I’m interested in taking this class because I am curious about how autonomous cars will change the physical and social aspects of cities. I’m interested in the potential interactions between AVs, public transit, density, and suburbanization.


Self-driving cars driving on city roads is an almost inevitable fact at this point — the technology is advancing quickly, and policies will eventually catch up. Policies are the real hurdle to overcome before self-driving cars can overtake human-driven cars as the main form of private vehicle transportation.

Policies will determine who or what dominates the roads in the future. Will the rules of the road be dictated by what makes traveling in autonomous cars most efficient? Will cars — after a century of domination — take a back seat to other forms of transportation, now that people are no longer at the wheel? Will there be a segregation of uses (autonomous cars dominating the transit zone, people and bikes dominating the social zone) like that seen in the woonerfs (i.e. home-zones) of Europe?

The introduction of self-driving cars onto city roads will likely have significant impacts on the way we conceptualize public transit (door-to-door minibuses?), who can be a “driver,” and many other topics that we have yet to think of.