The details are nebulous, but the more you read about autonomous vehicle (AV) technology, the more one message comes through loud and clear: The development and widespread use of AV is not an “if’,” but a “when,” and most importantly, a “how.”
The when of course depends on what you mean by autonomous vehicles. CEOs of automotive manufacturers and tech companies have been making predictions about when AVs will start rolling off the production line, starting with 2017 and stretching into the 2030s. In order to make sense of those predictions, it’s important to know exactly what they are referring to. There are varying degrees of autonomy, from cruise control and steering assistance to fully automated city driving. And there’s various contexts where these different degrees of AV could safely operate, from delineated zones on campuses or industrial sites, to dedicated lanes in cities, to… well, everywhere. Luckily, an international association of engineers, SAE, has come up with a handy breakdown of the six levels of driving automation:
Right now it’s safe to say SAE 2 is very commonplace (think cruise control) though there is plenty of room for improvement even at this basic level (think steering assist keeping you from a head-on collision with a minivan). SAE 3 is where a growing number of car manufacturers are at with their designs, where cars are equipped to sense what’s going on around them and essentially drive themselves, but human drivers must be present to handle anything that goes off book (and to take responsibility for the consequences). SAE 4 and 5 are where shit gets real, and AV technology stands a good chance of transforming our cities and towns in the same way the automobile itself did at the turn of the last century.
But even right now, at the current early stages of automation, our transportation planners and decision-makers should be paying attention.
Improved safety technologies, like lane departure warning and assistance, could actually help prevent head on collisions. Cadillac’s SuperCruise system released this summer includes a camera pointed at the driver to ensure their rapt attention to the road. After some warning, SuperCruise will actually turn on your hazard lights and slow down the car if you continue checking your phone, unwrapping your breakfast sandwich, or falling asleep, whatever it is that is keeping you from looking where you are going.
So this partial automation — the kind already available for the very wealthy and on the horizon for the rest of us – may be able to make driving safer. These improvements may just be a milestone on the industry’s race towards the fully-autonomous vehicle, but they could make quite an impact if they manage to reduce collisions, particularly head-on collisions, significantly.
And they will be an important development for the Nova Scotia government to take into account as it plans its next decade of infrastructure spending, or even as it considers actualizing the doubling of spending on highways promised by the Nova Scotia Liberals in their last election campaign. The cross-province tolling consultations and reports did not mention AV technology and what it might mean for highway demand over the new few decades, which the government estimated was around $2.4 billion. Since twinned highways in Nova Scotia are justified almost exclusively for safety reasons, these new vehicle technologies could free up huge amounts of provincial taxpayer dollars for other priorities.
According to Nova Scotia transportation and infrastructure renewal (NSTIR) spokesperson Brian Taylor, Nova Scotia has no formal take on AV. “Currently, NS does not have any policies or regulations related to automated vehicles,” says Taylor. “Staff are monitoring the development of this new technology to determine how best to incorporate them (sic) into the rules of the road.”
And most of Canada is in the same boat, just waiting and seeing what will happen with AV. Ontario is the standout, where the government is investing in AV research and development, and already has a pilot regulatory framework to test automated vehicles on its roads.
Maybe what NS Transportation and Infrastructure Renewal should be doing is offering up our rather imperfect, multi-seasonal roads for testing by some of these manufacturers, and funding some university research into ways that our road infrastructure can actually work with new vehicle technologies to make them even safer. (Snow-covered lane markings might not matter if vehicles have an alternate way to sense them.)
The full monty
Of course, fully autonomous vehicles, of SAE levels 4 and 5, will absolutely require a policy response and also demand infrastructure investments, and not just by our provincial governments, but by federal and municipal levels as well.
There are many somewhat rosy predictions of what the advent of fully autonomous cars could mean. Rosiest among those is that somehow this technology will push people towards seeing transportation almost exclusively as a service, and we will all own fewer personal vehicles as a result. Downtown parking spaces will become irrelevant, and be converted into residential, retail and recreational spaces for happy urban dwellers. Conventional public transit (depending on who you read) will transform to the perfect high-capacity, low-cost complement to fleets of shared autonomous vehicles available on demand. We will all save thousands of dollars in transportation costs, mostly associated with our shedding of private vehicle ownership. (And of course, over 500,000 people will be out of a job, the only real pea under the mattress in this optimistic projection.)
But will the simple advent of a technology, driven by industries whose endgame has and always will be maximizing profit, be all that is required for a best-case outcome like this? It seems highly unlikely. Why, for example, would manufacturers build for a world with fewer vehicles? The answer is they likely won’t, unless the rules make it so they must.
A discussion paper by researcher David Ticoll commissioned by the City of Toronto in 2015 posited three possible directions for things to go, post-full automation: privately-owned AVs will dominate, on-demand transportation (either public or private) will dominate, or there will be a split between the two.
A world where privately owned AV’s are the norm would reap few benefits in terms of reclamation of parking space and infrastructure, and even possibly exacerbate congestion if all those AVs end up doubling the kilometres travelled on a typical commute. And cheaper personal transportation, if that is even an outcome of AV, could incentivize sprawl. The advent of AV, in other words, is not some future cure-all that will solve our transportation problems. The roll-out of SAE levels 4 and 5 will almost certainly make transportation safer, but not necessarily more efficient or less detrimental to our air and climate. For that, we will rely on urban and transportation planning more than ever.
Halifax’s current Integrated Mobility Plan, our first multi-modal transportation plan ever, “acknowledges self-driving vehicles, but it is not part of any strategy that we are advancing in the IMP,” says Rod McPhail, the consultant who lead the 15-year transportation plan team. We’ll get to see for ourselves in November once the plan finally makes it to council (it was originally due this summer.)
McPhail says that AV will likely be more of a consideration in five years time when Halifax is updating its IMP. Considering that the IMP already represents a new way of thinking for our planners and decision makers, and considering AV technology is still developing mostly within SAE 3, waiting for the first review is fair. But let’s not kid ourselves, planning is about the future, and AV is definitely going to be part of ours.
There are experts in philosophy, law, and artificial intelligence who have suggested that the move to AVs will soon come to a screeching halt, due to the ethical and legal conundrums that will arise when computing and sensor technology reaches the point when AVs will be able to perform ethical calculus in emergency scenarios. There will be cases where smart cars will literally be asked to decide who lives and who dies. Imagine an oncoming vehicle suddenly veering into the lane of a smart car on a busy, pedestrian-lined street: the car has to decide whether to brake and head into the collision, veer into the oncoming lane and hit the car behind the stray vehicle, leave the road and hit pedestrians, etc.. There are even scenarios in which smart cars will have to decide which of their own passengers to save: e.g. depending on how a car steers into an unavoidable collision, front or rear seat passengers may be saved, but not always both. The calculus used to make these decisions will ultimately be approved (if not programmed by) humans. And who gets to decide which lives are most important?
It’s kind of freaky not having your hands on the wheel but having to keep them close. I tried it on some secondary highways set to specific speeds and didn’t like that too much. Some curves shouldn’t be taken at 80… It’s pretty amazing tech and definitely the way of the future.
My big objection to AV technology is that it perpetuates the single occupant vehicle culture when we should, as a society, be working together to eliminate the incentive for 1PVs. At the MacDonald bridge this morning, literally hundreds of 1PV’s trying to squeeze into the toll booths and buses filled with people being treated the same as the SUV with the loan occupant. The third lane on the bridge should be for busses and high occupancy vehicles to discourage all those people who are not otherwise given an incentive to use mass transit. AV technology will not change that except that the riders in their self driving cars will be on their smartphones texting.
If the “all AVs are privately owned” scenario comes true, it will still probably result in a reduction in the amount of space taken up by parking, and it will almost certainly change WHERE parking happens.
On the first point, cars that can park themselves don’t need to open their doors. They are also able to park with millimetre-level tolerances. So parking garages and lots can get WAY more space-efficient.
On the second point, a car that can drive itself doesn’t necessarily need to park next to its passenger’s destination. Why use valuable downtown real estate for parking when the car can drive itself to the edge of downtown and park on much cheaper land? Of course, there’s a trade-off with congestion from empty cars dead-heading to and from the outskirts, but I can see cities building centralized (and space-efficient) garages on less desirable land to serve each neighbourhood.