The Current - How Robotaxis could reshape Canadian cities

Episode Date: December 9, 2025

Waymo has set its sights on the Canadian market. The self-driving taxi company owned by Google parent company Alphabet, runs autonomous vehicle taxis in a number of American cities. Now it's exploring coming to Toronto. How safe are they? And can they handle winter conditions? And why some experts say self-driving cars will reshape cities and landscapes, the same way the car did one hundred years ago.

This ascent isn't for everyone. You need grit to climb this high this often. You've got to be an underdog that always over-delivers. You've got to be 6,500 hospital staff, 1,000 doctors, all doing so much with so little. You've got to be Scarborough. Defined by our uphill battle and always striving towards new heights. And you can help us keep climbing.

Donate at lovescarbro.cairbo. This is a CBC podcast. Hello, I'm Matt Galloway, and this is the current podcast. If you have never been in a Waymo, an autonomous driverless vehicle, your first trip can be a real trip. I'm not going to lie, I'm a little bit nervous about this, but let's go. Start ride. Hello, from Waymo. As we get going, just give us one minute to cover a few riding tips.

We'll do all the driving. Don't attach the steering wheel or pedals during your ride. So weird. The robotaxies owned by Google's parent company Alphabet are on the streets in the Bay Area, Los Angeles and Phoenix, with plans to expand to Miami, Washington, and London. And now Waymo has set its sights on the Canadian market. Last month, Waymo added itself to Toronto's City Hall's list of lobbyists asking to talk to city officials about the possibility of coming north.

Waymo, Tesla, Uber, they're all in a trillion. dollar robotaxie race to unleash fleets of self-driving cars in major cities. In a moment, we'll hear from somebody about how robotaxies are set to transform urban landscapes and economies. But first, I'm joined in our Toronto studio by Stephen Wasslander, director of the Toronto Robotics and AI Laboratory at the University of Toronto. Good morning. Good morning. I have been in a Waymo in Phoenix. You have ridden in a self-driving car before. Unnerving might be a word that is thrown around. How did it feel for you? Certainly at first it was unnerving. But yeah, it is an amazing experience. And the longer you're in it, the more you trust it, assuming everything goes according to plan.

Assuming everything goes to plan. But the unnerving part is that there's nobody there. Right. Absolutely. It does take a little while to get used to. But the technology has progressed to the point where this is really a viable business. This is something that's rolling out in cities and driving thousands of people. You believe that it's a viable business. I do. Yeah. It's hard to tell on the cost side whether this is really that much more efficient than having humans. because the cars are more expensive. There's more operational requirements in terms of keeping them clean. But it does seem to be growing and growing quickly. You surprised that they would consider coming to a Canadian market?

No, not at all. I think it's a natural progression. It's a big city. There's a solid taxi market here. So I see it as a natural step. For people who have never been in one of these vehicles, just explain how they work. So the car is equipped with additional sensors. There's cameras pointing in all directions.

There's a LIDAR which captures geometry or 360 degree point cloud information and tells you essentially where everything is. And then there's radar as well, which are really good at detecting vehicles, particularly at longer range. You get in the car, and as we heard, I mean, it says don't touch the steering wheel. There's nobody there in the driver's seat and away you go. How does it know where to go, but also how does it know not to plow into things on the way there? So both things are handled by the sensing and perception set. So essentially what's going on is the vehicle. is constantly monitoring everywhere around it for any kind of movement relative to a known map.

So it's pre-mapped all of the lanes, all of the traffic signals, and so it knows exactly what it should be doing in every instance in the map. And then it's looking for any dynamic or moving agents in the field and tracks those and monitors what they're going to do in the next five to ten seconds. Dynamic and moving agents being people or dogs or cats, cyclists, cars, cars, trucks, everything. Everything that might be going up against the static matter. And so this is really the big difference between the self-driving cars and the human drivers is that they have full 360 degrees situational awareness at all time. Where they falter is where we excel, right, is in the ability to use common sense and understand sort of complex and weird situations.

But every part of the day-to-day simple driving, right, the easy miles, those are fully mastered by this AI technology. What do we know about how safe Waymos are? Because one of the unnerving parts of this is getting in and closing the door. And again, we keep saying this, but there's nobody there. You're trusting the computer to get you around. In the same way that we trust aircraft on autopilot and the like, right? So we're used to this to some extent, you know.

But the safety case has been really impressive. And this is one of the nicest things, I think, about how Waymo's expanded. They've been very open about how they prove and confirm internally that they're vehicles are safe. They've published their own track record, so the miles driven and the number of incidents they've had. And they're exceeding human performance. They're actually in fewer accidents have caused fewer injuries and have no fatalities to date. We trust the maps. I mean, I live on a one-way street, and I would say almost every day an Uber goes down the wrong way on the one-way street, because the map has told it that the street goes the other way. Interesting. Yeah, that's

certainly an issue, but wherever this gets flagged, obviously the maps can get updated and the whole fleet benefits from it. So that's the advantage of having an autonomous vehicle doing these things, and that, you know, there will be obviously mistakes in the map and those need to get corrected, but as they're caught, every vehicle benefits. The mistakes get a lot of attention. Last week, Waymo announced that it was recalling its software after its vehicles repeatedly failed to stop or slow down for school buses in Texas. Waymo drove into an active police standoff in Los Angeles. There's video of this where there's a standoff there and the car just kind of turns in like it doesn't see what's going on. There was also the sad case

in San Francisco of Waymo running over Kit Kat, which is beloved neighborhood kitty that was there on the streets. The car didn't see it. I mean, what do those stories tell you about the technology and its limits? It tells you exactly where the hardest part of this challenge was and is, and that is what we call edge cases, these sort of rare events that don't happen very often and that aren't necessarily easy to understand, right? So the police standoff is an excellent example, right? This is not, there's no real signaling other than that there's cop cars in the way, there's police, you know, waving and gesticulating. You have to interpret all of that human behavior, right, which doesn't happen on a regular basis. But a human would

interpret. A human would say this is a police standoff. We can't drive in there. Exactly. And that's that common sense argument, right? So all of the technology needs to understand each one of these individual cases. Pets and small animals are extremely challenging to track and detect because they can move so quickly and have erratic behavior. And so that pushes actually the boundary of what's possible with the sensors they have available in the compute. And that's why that's a difficult challenge as well. So the reason that these are the stories coming out and not we hit a pedestrian walking across a street is that all of the primary use cases, all of the main events that happen in driving have now been solved. And they're now working really hard

on all of these final changes, right? The Texas example is a great one with the school buses in that those school buses looked a little bit different. The rules didn't quite match what the AI agents in the cars were used to, and so they were making the wrong choices. So Waymo immediately reacts when they get these reports, they update the software, and that's no longer an issue. So within a week, they were able to turn that around, from what I understand.

Phoenix, Los Angeles, San Francisco, there's no snow there. I mean, they might occasionally get a little bit of snow, but certainly not what we get here. Would they actually work in winter conditions? So it's been exciting. I've been working on the winter autonomous driving problem for the last five years now. You and I have spoken about this before. Absolutely, yeah. And what we found was that it certainly has an impact, right?

We all know that winter driving is harder and different. And that's sort of the way the cars treated as well, the autonomous cars. So they need to learn how to see through the snow in the vision data and in the LIDAR data. They need to know what behaviors are changing, so how people act differently, how they cross streets slower, you know, when they get stuck in snow banks with their vehicles, et cetera, right? Lots of different behaviors are happening, and they need to know the limits of their own traction. So how carefully they need to proceed when they're on slippery surfaces, right? All of this stuff has to get included into that pipeline of software that's controlling the vehicle. And that's possible?

It is. That seems like a lot of challenges. I mean, some of us, humans struggle with some of those issues. So you reduce your speed when you're in difficult situations in the winter, right? And you manage the risk, essentially. And so that's the same thing that the Waymo vehicles and the autonomous vehicles have to handle. They have to be able to understand that the same speeds that they would normally operate are no longer viable. They need to, you know, dial back their predictions of what people will do and therefore, you know, basically adapt to winter driving.

So it's not an impossible challenge. It's just more work. What are the things that you want cities to think of? about before these things arrive. We're in the city of Toronto. There was this pilot project that was done by Magna International, this car parts company,

where they had this self-driving delivery vehicle, kind of looked like a suitcase on wheels that was rolling through some neighborhoods in the west end of town. It didn't go particularly well because it would veer out of its lane. It would stop in the middle of the lane and kind of reboot itself,

have this kind of unscheduled shutdown. Apparently wasn't paying attention to some of the rules of the road. And there were people who were saying, see, it's not fit for purpose. This thing doesn't actually work. There are a lot of questions around what data they gathered, but it was something that was actually happening on the streets

of a major urban center. So what would you want city officials, if they're now being lobbied by Waymo, to think about before these cars end up on our Canadian streets? I think the most important thing is to ensure that there's a clear regulatory environment and hopefully a national one, one that requires companies to report all of their performance measures so that we have full clarity on when the failures are happening, what kind of mistakes they're making, and what they're doing to prevent them in the future. I think we need to understand how these companies can then progress from one city to the next and provide this service more broadly, and they can't just happen sort of willy-nilly. We can't have one city at a time,

you know, figuring out what they think is right and then the next city coming up with another example. What's at stake if some of those big questions aren't answered? I think it'll just delay the progress. It'll delay the opportunity for self-driving to come to Canada if we don't figure these things out now. And you trust the companies to play by the rules? I mean, again, what we're going to talk about in just a moment is when ride sharing arrived in many cities. Right. There were rules.

They just ran over the rules. And then cities had to figure it how to adapt after the fact. Yeah. I think it's harder to stop the right, or it was harder to stop the ride sharing. It was more of an organic growth that exploded through the app market. Here you have, you know, a licensed vehicle that can't simply. be, you know, put on the roads without, without government supervision. So I think it's an easier to control market because it's going to be robotaxy fleets as opposed to individual

users. Last question. And again, you and I have spoken at this before. The joke with, it's a different industry, but people talk about fusion, nuclear fusion. It's 10 years away from being 10 years away. Right. 20 years away from being 20 years away. Is this close? Is this a real thing? It's a real thing. Absolutely. And, you know, you've driven them yourself. And people in San Francisco now use it by default. Many of my colleagues that I've talked to find that it's more convenient, easier, and as it turns out, it's actually safer. So the technology is here. Ten years ago, I was asked when it would come. And I said, in ten years, hopefully, and I think I was pretty close to the mark there.

Stephen, thank you. Thanks for having me. Stephen Wesslander is director of the Toronto Robotics and AI Laboratory at the University of Toronto. This ascent isn't for everyone. grit to climb this high this often. You've got to be an underdog that always overdelivers. You've got to be 6,500

hospital staff, 1,000 doctors all doing so much with so little. You've got to be Scarborough. Defined by our uphill battle and always striving towards new heights. And you can help us keep climbing. Donate at lovescarbro.cairot.ca. Are your pipes ready

for a deep freeze? You can take action to help protect your home from extreme weather. Discover prevention tips that can help you be climate ready at keep it intact.ca. That robotaxy race is heating up. Uber says it's planning to scale up its fleet to 100,000 self-driving cars. The company's CEO, Dara Kazashai, is predicting big changes to how we travel. I think all cars being autonomous, I say 20 plus years. And I think that driving will be something like horse.

horseback riding. Like it's really fun. It's a way to get around, but it's something that you do maybe in your leisure time. David Zipper is a senior fellow at the MIT Mobility Initiative. He looks at how transportation and technology overlap. David, good morning to you. Good morning. Thanks for having me. Thanks for being here. Tech optimism is not uncommon in Silicon Valley. What do you make of the CEO of Uber's prediction that driving is going to be like horseback riding? Well, it's a great line. And I feel like I've heard that before many, many, many, many times over the last 15 years. There have been all kinds of predictions about autonomous driving being widespread by 2018, by 2020, by 2025, and looks like now it's 2045, and I would

suggest taking those sorts of estimates with a grain of salt. You've also written, though, that the rise of self-driving cars promises to bring the most tumultuous shift in transportation since cars first rumbled their way onto the scene. So how is self-driving going to reshape the cities that we're in right now, if it actually happens? Yeah. And so if and when it does happen, I do think it will be a major shift. But that shift may still be a few decades away.

I think it's really important to keep in mind that there's a world of difference between having a city with a few dozen or a couple hundred self-driving cars and having a city with, you know, 50,000, 100. thousand plus. And that's really, I think, what the self-driving car companies are aiming toward when they get there is, I think, still up for debate. But when you have a city with 50,000, 100,000 self-driving cars going around, it really transforms, I think, the transportation network. And in part, I think, it's going to lead quite simply to a lot of congestion. You said for a couple of, well, you continue. I mean, part of the couple reasons. One is that you

compare it to what happened with ride share, right? Yeah, yeah. I mean, it's kind of funny. You know, I'm old enough to remember that, you know, 15 years ago, there was a lot of talk about how Ride Hill was going to liberate us from the need to need to own a car and, you know, your car sits unused 95% of the time. There'd be less driving. None of that actually happened. According to research that's been done now, when Ride Hill emerged in cities, you actually saw an increase in total car registrations and an increase in total driving. And an increase in total driving. And an increase. in congestion and a decline in transit use. That's been repeatedly documented. So I would expect if and when robo-taxies become widespread, you're going to see people take more trips because it's a nice way to travel with the privacy and you control the climate. It's a very consistent experience being inside them.

And then there's one other element, which I think is important to keep in mind that Robotaxies and Ride Hill both have. And that's called deadheading. Those are the miles or kilometers that are driven with nobody being transported anywhere because the vehicle is en route for a pickup or it's waiting to be summoned. And with ride hail, it's about, you know, third give or take of total driving is empty or deadheading. And so far in California, with Waymo Robotaxies, about 50% of all the driving is empty. It's deadheading.

So if you think about that at scale, you're adding on a lot of sort of empty calories of driving. onto a highway network, and as best I know about Toronto, there's not a lot of space to expand the highway network, so you could see a lot more congestion. One of the things that you say as an opportunity in light of all of that is around parking and our relationship with parking. Just briefly talk about how cities might be able to rethink parking if these things arrive at scale. Yeah, well, that actually is one of the possible upsides of robotaxies from sort of like an urban planning or an urban policy perspective because robotaxies don't really need to park upon

arrival. They simply need to stop and do a pickup or a drop off. And that being the case, there's an opportunity potentially, and I don't want to oversell this, but potentially to repurpose some of the on-street parking that lines the streets in a city like Toronto and most cities across North America. Perhaps some of that could be turned into parklets, like little miniature parks or maybe bike share stations or whatever else it might be. That could be very positive. Do we worry about job losses? Again, we saw this with ride hailing, that a lot of taxi drivers ended up losing their jobs when companies like Uber entered this in the scene. Would something like this lead to significant job losses, do you think? I mean, quite potentially. That's been

a topic of considerable debate already in the city of Boston, Massachusetts. There is big pushback against Waymo's interest in deploying in that city. And the opposition is really coming from labor, from organized labor taxis, as well as from those who work for ride hail, Uber and Lyft, who worry, I think with some, I can understand where they're coming from. They worry that if and when robo taxis really scale, they'll put them out of a job. So I think that there is a concern there. And so what do you want municipal leaders in Canadian cities to do to get ahead of some of these issues. Again, one of the things that happened when ride hailing companies arrived, many people believe, is that municipal leaders were caught on the back foot and the tail

ended up wagging the dog. That's right. That's exactly right. The same thing happened, by the way, Matt, in the United States. I think it makes a lot of sense now for cities like Toronto to be thinking about not just what would it mean to have our first Waymo's or robo-taxies deployed in the city, but what might Toronto look like? with 100,000, 200,000 of these vehicles on our streets. And are there policies that we might want to think about developing now or putting in place now to sort of inoculate our transportation network from some of the potential risks, which includes congestion.

It also includes potential, just sort of levels of chaos of sorts. Because while it's very impressive the technical achievements of robotaxies, in a city like San Francisco, there's been, countless documentations of Waymos and other robotaxies that have broken a whole variety of traffic laws, taking illegal turns, picking people up or dropping them off in a bike lane, blocking transit lanes. And if you, again, if you scale that by 100x, the number of robotaxies, that could create real problems for an overall transportation network. So to answer your question, I think it's a great time to think about something like congestion pricing, which can mitigate the risks

of standstill gridlock. It's also a great time to be thinking about automatic enforcement, which I recognize as a controversial topic in Ontario, but it is a good way of sort of trying to train or push robot taxi companies to avoid violating traffic laws, even if the excuse could be well,

you know, humans do it too. I have to let you go, but your sense is that those are the conversations that need to happen right now. I mean, it might be 10 or 20 years down the line, but this technology is developing at pace. Yeah, you know, like, Darik Haas Rshari says it's coming soon.

Maybe he's right. There have been a lot of predictions made of all kinds. But if and when robotaxies really do scale, I think we'll be really glad if we think now about the kinds of policies we should be enacting to make sure we don't regret it. David, really good to talk to you about this. Thank you very much. Thank you. David Zipper is a senior fellow at the MIT Mobility Initiative. He looks at how transportation policy and technology overlap.

You've been listening to the current podcast. My name is Matt Galloway. Thanks for listening. I'll talk to you soon. For more CBC podcasts, go to cbc.ca.ca slash podcasts.