Experts say autonomous vehicles could solve some tricky commuting problems — but there’s still a long road ahead for the technology
By Josh Sherman - Published on Oct 04, 2021
What some say could be the future of public transportation in Whitby has a top speed of 20 kilometres an hour.
For safety reasons, that’s as fast as the GTA suburb’s new eight-seater driverless vehicles will be travelling when they start transporting their first passengers along a six-kilometre route, said to be the longest AV circuit in Canada to date, in the next two or three weeks. “The future’s coming — very slowly,” laughs Jamie Austin, deputy general manager of business services at Durham Region Transit, who says testing for the shuttle began last month. The pilot is expected to last until late 2021 or early 2022.
The driverless vehicles come equipped with GPS, cameras, radar, and lidar (short for light detection and ranging). Computers process the information fed from these devices as the AVs navigate pre-programmed routes, although on-board operators will be present to take the wheel if neeIn Ontario, the idea that micro-sized autonomous vehicles could be an answer to this so-called first-mile/last-mile problem is gaining traction with both researchers and municipalities. But don’t expect to see major changes anytime soon: experts say that although technology is evolving, a number of significant short- and long-term challenges — from the complexity of roadways to some daunting economics — mean a computer-guided commute may still be a ways off.
Amir Khajepour, an engineering professor and researcher with the University of Waterloo’s Autonomous Vehicle Research and Intelligence Lab, says self-driving vehicles could save municipalities money down the road by allowing them to cut staff expenses — and potentiallyby operating on-demand, rather than continuously. “If the system is fully autonomous, as soon as there is a huge surge in terms of the number of passengers, you can run five of them right away, continuously, and then, as soon as it drops, you park two of them,” he says. “This is not something that you can do with a full-driver system, because what are you going to do with the other two drivers? They come for two hours, and then they go home?” But, for now, cost is a barrier. “Obviously these buses are more expensive,” says Khajepour.ded.
The hope is that these small vehicles will help solve a problem that transit authorities across North America are facing: how to connect transit hubs — such as GO stations — with nearby areas, often suburban or rural, when traditional buses prove too costly for lower-volume ridership. “That’s often the biggest challenge, just getting people to our service,” says Austin.
Amir Khajepour (fifth from left) is an engineering professor and researcher with the University of Waterloo’s Autonomous Vehicle Research and Intelligence Lab. (Courtesy of Amir Khajepour)
According to Tenille Houston, co-founder and CEO of AutoGuardian — which is providing technology for the Whitby pilot and for a similar one in Toronto that’s expected to be in operation by November — most municipalities, including Toronto and Whitby, are leasing the compact shuttles, which run in the ballpark of several hundred thousand dollars to purchase. “As technology evolves, the cost is expected to decrease,” she adds.
Another thing yet to be determined: how long these vehicles will remain operational. “What is not understood at this point given their infancy is the life expectancy of a typical autonomous shuttle (they just haven’t been around that long) compared to a full size bus which will typically last 12 to 15 years (with a mid-life refurbishment of the engine and transmission),” Austin tells TVO.org via email. “One of the benefits of piloting these types of technologies is to get real world data on operating costs that can inform the business case for the vehicles. That is, whether they can be a reliable service option while providing good overall value compared to alternatives.”
Winter performance is a focus of Toronto’s pilot, which is slated to run along a 4.2-kilometre route through February in the West Rouge community. “One of our research questions is really how much through the winter can the shuttle operate,” says Ryan Lanyon, chair of the city’s automated-vehicles working group.
With testing ramping up, how long before Ontario sees a fully autonomous shuttle operating as part of a transit network? “I think that might be a little further, especially without an attendant on board,” says Houston, estimating a timeline of anywhere from 10 to 20 years. “The challenge now is in the nuance,” she explains: AVs must respond to everything from local wildlife to extreme weather, so programmers regularly adjust the algorithms guiding the vehicles so that they react to different environments appropriately.
Steven Waslander, director of the Toronto Robotics and AI Laboratory at the University of Toronto, says that “it’s literally trying to identify all of the objects in the environment around it, track its own motion relative to those objects, predict what those objects will do next based on experience from similar objects, and then make safe plans for forward progress as well as emergency escape, and then execute those manoeuvres through the environment following road rules and safety limits for the vehicle.”
For the past two years, Khajepour has been heading Waterloo’s WATonoBus project, through which driverless vehicles are being developed and tested on the university campus’s winding, 2.7-kilometre Ring Road. One of Khajepour’s goals is to develop a safe and affordable autonomous-vehicle system that can then help municipalities establish their own. A more immediate aim: providing on-campus rides at Waterloo. “When you have a shuttle like this, there are a lot of unexpected things happening in a busy environment; you’re not on a highway where the only thing you see are trucks,” he explains, noting his team has had to grapple with challenges such as geese wandering into traffic.
Research and development is only one of the areas in which much remains to be worked out. “There are many things that obviously need to be addressed,” says Khajepour, citing regulation and insurance policies. “The other thing is related to some of the ethical aspects,” he adds. In a serious collision — for example, one in which an autonomous vehicle struck and killed a pedestrian — who would bear the responsibility? “Who are you going to charge? The car? The company?” Khajepour asks.
An autonomous-vehicle pilot is set to be up and running in Toronto by November. (Coutesy of the City of Toronto)
Late last month, the Ministry of Transportation announced proposed changes to the 10-year pilot program that has made current AV testing possible on public roads in the province. The most beneficial change for industry, says Houston, would be allowing AVs approved under the pilot to be part of the province’s Manufacturer Plate Program. “We currently are under special permit to operate,” she says via email. “This is a permit that must be renewed every two weeks for every vehicle and as we have three vehicles on the road for our upcoming deployments, that would take a lot of time.” The province says in a news release that the program accomplishes multiple goals: “The pilot framework ensures that the province’s roads remain safe while also supporting economic growth and innovation.”
To date, says Waslander, nobody has been able to prove that AVs can operate as safely as human-driven vehicles, despite massive financial investment. “They can be made very safe,” he adds. “But then they don’t make a lot progress, and they tend to be overly hesitant and frustrating to be around.”
Waslander says “it’s hard to put a number on it” when comparing the safety of AVs to human drivers. But he points to research out of California that suggests driverless cars can go 25,000 kilometres without a disengagement — “which just means the software wasn’t sure what to do so it stopped or requested a [human] takeover” — whereas a typical human driver has a fender-bender every 500,000 kilometres. “Proving that you can do it as reliably as a human for every vehicle you put on the road so that you no longer have to watch it — that’s where we’re all stuck.”
Although the tech is clearly not ready for widespread adoption yet, he says, the way in which Toronto and Whitby are piloting AVs does decrease risks. “For restricted routes or for very low-speed operation in uncluttered areas, they can be almost as reliable as humans or as much so.” At 20 kilometres an hour, he notes, a vehicle can stop on a dime, compared to, say, 50 kilometres an hour, which may require up to 40 metres stopping distance.
And then there’s figuring out how to account and adjust for passenger behaviour. Terry Johnson, a spokesperson for Transport Action Canada, suggests that AV shuttles could indeed solve the last-mile dilemma and thereby improve transit accessibility overall. However, he, too, has longer-term concerns. “Removing a professional driver from the vehicle is not necessarily a good thing, and so we really want municipalities not to think of AVs as a way to cut jobs and save money. Bullies tend to be emboldened in small group, unsupervised situations,” he says, adding that cameras won’t be enough for public confidence.
Read the full article here