AI can help them run their agencies more efficiently, um, but it, it, still needs their input. Stop Requested.
This is Stop Requested. by ETA. I’m Christian. And I’m Levi. These are real conversations with the innovators, operators, and advocates driving improvements in public transportation.
Over the last two episodes, we’ve been broadly looking at how transit agencies are approaching AI. This week, we’re getting more specific. Our guest is Charlie Territo,
Chief Growth Officer at Hayden AI. Hayden uses vision AI on transit vehicles to identify bus lane and bus stop obstructions, understand what is slowing down service, and give agencies new data about what is happening on the street. We talk about how that technology works, how automated enforcement can change driver behavior, what agencies can learn from the data beyond enforcement, and where vision AI could go next.
Here’s a. conversation with Charlie Territo. Welcome back to Stop Requested. Uh, we’re in for another great show today. Uh, we’ have
Mr. Charlie Territo, Chief Growth Officer for Hayden AI. joining us this morning. Uh, Charlie, how you doing this morning? I’m doing great. Thank you so much for having me. Of course. And, and, and you guys are, um, uh, trending in, you know, the application of AI.
I think if, uh, organizations around the world, across industries, if they’re not using AI. or really thinking how that is going to be implemented in their businesses and their practices, they’re gonna be left behind.
Uh, so Charlie, could you please introduce yourself? Tell us a little bit of this, uh, role that you have as Chief Growth Officer at Hayden AI. Sure. Uh, my name is, again, Charlie
Territo. Uh, I’m the, the Chief Growth Officer for Hayden AI. And, and what that means is that, um, I focus on, um, our, uh, business development.
Um, my teams focus on our customer success or, or management of, uh, new customers as they come on board. Uh, my team focuses on, um, helping our customers grow and expand the use of our systems. Um, and then also the, the marketing and communications, um, that we use to support our programs.
Oh, wow. Tho- those are a lot of hats. And, and, and let me ask you, do you guys focus mostly on the transit industry or also across industries? You know, right now our, our core focus is transit.
Um, um, uh, you know, almost all of our current deployments, um, are on transit buses with the exception of, um, a, a few emerging uses of the technology that we’re exploring in, um, some of the, the cities that we’re working in.
But for the most part, um, our systems are really optimized, um, for, um, transit. And, and I wanna talk to, uh, to you a little bit about your career. So y- you work in government, the auto industry, automated traffic enforcement, a- and you’re now in public transit.
Uh, what, what, um, pull you into the transportation industry in the first place? And could you tell us a little bit of that progression of your career?
Yeah, absolutely. You know, I, I think one of the, the things about, um, mobility is that it’s so ubiquitous.
Um, everywhere we go, um, cities are, are struggling with the challenges of moving people from point A to point B. And whether that’s with, um, cars or whether that’s with subways or whether that’s with transit buses or, uh, micro mobility or shared mobility, um, it- it’s an area that, um, it has been a very important part of, of how cities move.
And, you know, I, I think for me, um, o- of all the different, um, elements of, uh, o- of a, a- an urban environment, transportation’s the one that I always found, uh, the most interesting.
Um, it, it, it, a ma- I remember years ago watching a show about parking and, uh, a show about traffic and how parking impacts traffic.
And to me, that was somewhat fascinating, um, how, how one vehicle parked, um, in a certain, um, s- place, uh, but illegally, uh, could have an impact on an entire traffic network.
And so, um, I, I think, uh, uh, it wa- it was those types of experiences, um, that kinda gravitated me towards transportation.
Uh, could you tell us, uh, first of all, I’m curious about the show. Do you recall the show name? Because now I, I wanna see it. I don’t recall the show name, but I can, uh, I can think about that. I, I think it was on, um, like a, a PBS, like a Frontline type show.
But, um, but I, I mean, you know, it, it was years ago, so. Okay. Uh, yeah, I mean, uh, uh, it, it sounds interesting. And, and to your point, uh, you know, the city, any city is, is, i- in, you know, as it relates to, uh, transportation, is an ever-changing environment. Like, there’s so many factors that are impacting it. It’s a whole ecosystem. So it’s always very interesting, uh, to be part of it and, and just see how, you know, different factors have that, uh, ripple effect on- Yeah … everything else that, that is, uh, part of that, uh, ecosystem.
So, y- you know, you joined, uh, Hayden AI in, uh, 2022. Uh, w- what did you see about the company, what was doing and happening that, that kinda pulled you into, um, you know, being part of it?
You know, I, I think that, um, one of the things that attracted me most to Hayden AI, um, was the technology itself, and, um, the fact that it was designed, uh, to serve what was, uh, uh, from my perspective, um, a very underserved part of the transportation network.
Um, there were any number of fixed camera systems, um, that were being used, um, to try and, um, you know, enhance transit efficiency.
Um, but there really hadn’t been, um, I think, a successfully deployed mobile-based system. And, um, the more I learned about the technology, uh, the more I learned about, um, how it could be deployed, and the more
I, I thought about its effectiveness, um, and its ability to really help transform the way large city agencies deliver their service, um,
I, I, I w- was interested and, and thought it was something that, um, would be a, a great opportunity to try and, um, try and grow. So not even five minutes in and we get a PBS
Frontline reference. I h- I, I’m really digging this episode so far , Charlie. Uh, that’s, uh, I, I gotta, I gotta check that out if that, if that’s what it was. Uh, you know, speaking of the, the technology, uh, and what, what it does, how it’s different from others, uh, and then also, you know, this whole parking, uh, conundrum, right? We get into areas where there’s, there’s n- not a lot of parking, and people park in places that they shouldn’t.
Uh, what is Hayden doing, uh, to, you know, identify where that parking behavior is i- interrupting bus service, or train service perhaps? Yeah, you know, I, I think this is, uh, a, an area where data can be very, very valuable, and there are any number of systems on a, a bus that, um, tell the bus, um, or tell the, the agency where the bus is and how fast it’s moving. The one thing those agencies don’t always have access to is the why.
Um, and with, with our vision AI technology, um, we can do both. Um, we also have, um, very robust lane-level accuracy for the data we collect. And so we can say, not only is a vehicle slowing down at this, this location, uh, this is the lane that it is where it’s moving slower or faster than you expected it to move, and this is why we believe that’s happening.
And, um, when you start to compile that data over time, um, you really can pre- present agencies with, um, data that previously wasn’t available, and that gives them greater insight into ways that they can enhance their service.
Um, you know, I always tell our, our customers that this is a three-legged stool. Um, one element, uh, is going to be enforcement. Um, the other element is going to be education, and the last leg of the stool will be engineering.
And there are going to be things that our data helps uncover that have nothing to do with, um, the, the way drivers drive, and everything to do with the way that certain segments of the roadway are engineered. And potentially by making some changes to, to engineering, or as I said, with education, posting certain signage, um, we can be just as effective as we are with enforcement.
So, um, uh, probably from agencies, th- they understand immediately why this is a problem. But t- to those who m- might not m- you know, see this as a prob- problem, they, they might see that, well, it’s a traffic violation, sure, but is it really that big of a deal? W- what do you say to those people? You know, I, I think that, um, a, a bus, um, is typically, um, traveling with not just one driver, but multiple passengers, and those buses are not designed to be able to maneuver in and out of traffic as easily as cars can or other vehicles. And so when a, a, a bus is in a, a dedicated lane, and that lane is blocked, um, for the bus to go around that vehicle, um, it is a, a, a, a challenge.
And, and that’s a challenge n- not just for the bus, but for drivers that then have to yield to that bus, uh, that’s trying to, to move in and out of those dedicated lanes. And so the, the impact of an illegally parked vehicle in a transit lane, um, is much more significant than an illegally parked vehicle at a metered space, for example. Um, because that lane is designed, um, specifically to enhance the flow of transit, and if that vehicle now has to move to another lane, the entire network is impacted.
Yeah. Uh, and I, I would agree with that, and I, I think that’s what, uh, you know, a, a lot of transit agencies experience, um, especially those with the dedicated bus lanes.
But on the, the bus stop front Uh, is that treated any differently? Like here in Florida, we have, you know, dedicated bus bays where, you know, maybe it’s an arterial road and pretty high speed, so you wanna get the bus off the road but still pick up passengers because, you know, it may be connected to some pedestrian network. Uh, are, are those treated differently with Hayden or, um, pretty much the, the same concept, uh, applied- Yeah … you know, equally?
Yeah, pretty much the same concept. I think that the, the difference being, um, that oftentimes a block- a blocked bus stop, um, is not as much, um, an issue for the efficiency of the bus, but more an issue of safety.
When, um, individuals, specifically disabled individuals, um, need to access the bus and are expecting to do so through, um, a, a, a curb cut or, um, some other, um, um, accessibility design, and that isn’t available to them, um, it creates a, a safety hazard. When a, a mother’s pushing a stroller and has to walk out into the street.
or around a parked vehicle in order to, to board a bus, creates a safety hazard.
Um, when, when someone is, um, maybe visually impaired and is expecting to board a bus at a certain location and can’t do so because that location is blocked, it creates a safety hazard. And so, um, typically, I think the, the, the way we identify the events are very similar, um, but the impact of the event on the network, um, is a little bit different in that, um, blocked bus stops really are an issue of safety, um, and less a- an issue of accessibility or of efficiency in some places. I see.
Yeah. And, and that makes sense. Uh, so from Hayden’s perspective, you all are, uh, providing that technology to be able to identify the objects that are, you know, perhaps blocking a bus lane. Is that right? Yeah. I mean, our- It-
Our, um, system is, is both, I think, very complex and, and very simple. Um, at its core, um, we’re going to build a map.
We’re going to annotate that map with what’s important. We’re going to train our systems of the way the map should look and what to do if it looks differently. And then, um, we’re going to deploy, um, that map to all of the vehicles that travel on those roads. And so, um, we’ll, we’ll tell our system, “This is a bus lane. You should do this if a vehicle’s parked here. Um, this is a bus stop.
You should do this if a vehicle’s parked here.” And, um, a- and, and any number of events or, or, um, types of violations that may impact transit, we can do the same thing. Um, for example, in
New York, we’ve trained our systems to say, “This is double parking, and if someone’s double parked, um, then this is what you should do.” Um, we’ve trained our systems in, uh, California to say, “This is a bike lane, and if someone’s parked in this bike lane, this is what you should do.”
So the, the system is, is actually very complex, but, um, if you think about it, you know, there are any number, any number of different maps that you may use. We build our own maps that, that tell the system what’s important on the transit network, what it should be looking for, and what it should do when it sees something that isn’t the way it should be seen.
Today’s episode is brought to you by ETA. Transit agencies rely on dozens of systems to keep service moving, but the information inside them often remains disconnected.
Transit OS connects operational technology, enterprise software, and public data so agencies can bring more of their information into a shared operational context. Transit GPT gives transit professionals a practical way to use that connected data, asking questions in plain language and getting answers grounded in agency information.
Together, they help agencies see service clearly, respond faster, and deliver more reliable trips for riders. Learn more at etatransit.com.
So it’s identifying those anomalies. Um, th- that’s really interesting. I, I appreciate that explanation. Mm-hmm. Does, uh, does
Hayden play any part in the enforcement education engineering sort of three-legged stool that you mentioned earlier? I wouldn’t think so, but, uh, I, I feel like it’s important, uh, to ask if there’s, y- you know, if you have your, your hand in, in any of those areas as well. You know, I, I would say tangentially we do. Um, we certainly interface with, um, transit planners and others that, um, oftentimes ask us for, um, data that can be used or visualization of data that can be used, um, that provides them with insights that previously they didn’t have.
Um, you know, uh, oftentimes, um, agencies know that there are, um, impacts of illegal parking on their network. But don’t quite know what the, the impact of that is. And so we’re able to provide that to them, we’re able to correlate that for them, and, and we’re able to even give them analysis like time of day and day of week when those occur the most often that can be used to make decisions about changes to the network. Um, so, you know, that, I think on the engineering side, um, we’re not doing engineering, but we are providing data that can be used to do engineering, and that’s very valuable. Um, on the, the education side, um, you know, we certainly work with our customers to share best practices from other transit agencies.
Um, we take, um, a- and have teams that can assist with, um, education efforts. Um, there are certainly some agencies that, um, re- rely on Hayden, um, to help, um, build certain, um, analysis that can be used in their communication outreaches.
And from an education standpoint, um, we’re willing to, to do what we can do, um, to b- to make sure that any agency looking to deploy an automated, uh, camera enforcement program, um, has access, uh, to other cities and agencies around the country that are doing the same thing that they can use as a, um, a, a, as a way to, to learn, um, some of the dos and don’ts. So, so you facilitate the benchmarking in a way? Like you help agencies talk to one another that have similar, uh, you know, camera enforcement programs?
A- absolutely. I mean, I, I think one of the values, um, of Hayden AI is that, um, currently we’re in five of the, the 10 largest transit agencies in the US, um, and we’re, we’re adding more every single day.
And those, um, those relationships are ones that we can connect, um, those agencies with each other, um, to learn a little bit more about what’s working, um, what, what they could be doing better, um, a- and ultimately, um, how they could be, should be thinking about, um, growing these types of programs in the future. Ha- have you had any, uh, aha moment for one of ’em where they spoke with somebody else and they said, “Oh, wow, like they’re doing this and that and it’s succeeding. We wanna do the same”? Like a- any, any knowledge transfer that, uh, translated into making a change? You know, I,
I think that the, the biggest, um, i- issue is just the ability to share their experiences with each other.
Um, I, I’m not, I f- I’m not sure that there are any yet that have kind of, um, relied on something that another agency is doing to change the way their program works, but I do think that there is a great value in sharing, um, what it is each agency is doing. You know, the, the reality of, of our business and of the automated, um, camera enforcement industry is that there really is a, a patchwork quilt of rules and regulations. And so what a, an agency in California can do is different than what an agency in Chicago, uh, can do, or an agency in New York can do, or an agency in, in Washington DC can do, or an agency in London for that matter can do. Mm-hmm. And so, um, it, it’s not always transferable. Um, but m-
I think one of the, the values is, um, just for them to see how it is other agencies talk about their programs, how it is other agencies promote their programs, and how it is other agencies grow their programs over time. Hmm.
Very interesting. And, and, you know, in a way, uh, um, as I thinking about what you’re saying, um, you know, and these programs are growing and when it comes to a enforcement is something that might not be very popular, right? Like i- if it’s you getting the ticket or, you know, like you like to park there or, you know, like use the space that is not supposed to be used by you. Uh, but in a way that, that is kinda like, um, education, right? Like, uh, you know, helping changing, uh, driver behavior, because the goal is not just to. enforce and issue tickets, right? Yeah. I mean, I,
I think, um, the hope and expectation is actually over time, um, the, the number of tickets will fall, um, driver behavior will change and, um, ultimately buses will move faster, um, more on time and, um, safer.
And if we can do those three things then our bet is that we’ll get more riders. And I think that’s the most, um, important element and the goal of every automated enforcement program is that at the end of the day, this results in more people deciding to choose transit because buses are faster, buses are more reliable, buses are safer, um, and, a- and it, it makes it more attractive for people to ride
So, so before having an automated, uh, system that, you know, AI cameras that are detecting and, and doing these things, would you say that agencies, uh, suffer of, like, non-enforcement?
So meaning, like, people realize they can get away with doing these things, and they do it- Yeah … and then the buses are being held up and, you know, tickets are not being issued. Like, you know, people are not being, um, you know, issued a ticket for, for, you know, breaking the law, right? Like, i-is that the reality? And, and do you see that, like, once you’ve put your system in place at these different transit systems, like in fact, first it was like a, a, an uptick of, you know, issuing tickets, like finding all these infractions, and being able to show the agency, like, all these instances that is causing, like, delays or obstructions to the, you know, transit system.
And then, you know, after a few months seeing the decline of, uh, infractions and then the improvement of, uh, on time performance or, you know, like the, the, the, the, the speed at which the bus moves, is that something that… You know, is that, is that a good description of what happens or what agencies experience after they put Hayden AI in place?
Yeah. I think that that is, uh, I think that’s what we hope happens. Um, you know, the, as, as we’ve said, traffic networks are very complicated.
Right. And so sometimes it takes time, um, or there are other factors that impact a, um, you know, a, a, a bus route or the ability of, uh, of, uh, the program to be as effective as it can be. Um, for example, um, the number of buses on a route, um, has, uh, a lot to do with, um, the rate of, of behavior change.
Um, if there aren’t a lot of, of buses on the route, if the interval from when a, a bus with a camera, uh, drives down, uh, a bus route is every six hours, then you’re probably not gonna get as much behavior change as if the interval of buses with camera enforcement driving down the, the route is every five minutes. And so there are any number of different fact- and that’s just one factor. There are any number of factors that can be, um, that can, can e-either accelerate or, um, slow down, um, the rate of, of behavior change.
Um, one thing that, that I think is, is interesting is that w-without automated enforcement, whether it’s fixed or mobile, um, a-an, an officer or a parking enforcement officer, um, giving someone a violation for parking in a bus lane, um, is going to, to typically create a violation themselves.
Um, a vehicle, you know, police officer that, um, sees someone parked illegally and parks their vehicle behind that vehicle to issue them a violation, um, is going to, to, to make, um, an additional obstruction.
Um, one benefit of a, a Hayden AI system, um, is that it’s only capturing violations when the vehicle in question is blocking the bus.
And so it doesn’t create an additional blockage, um, by enforcing the violation.
Um, if the vehicle is in front of the, the bus with a Hayden AI system, it’s going to capture that event, um, it’s going to create an evidence package, and it’s going to continue with its route. And so I, I think that there is a, a great benefit, uh, to having a system that can capture a violation without creating another obstruction for buses, um, on that route.
Oh, wow. That, that, that’s interesting. You know, I was not thinking about that. That’s true. You know, that, that, uh, police o- police officer, you know, law enforcement, they, they have to stop behind that person. Yeah. They do an entire interaction and asking for the driver license and this and that. But, uh, also, uh, people that are getting infractions are when they’re actually blocking the bus, right? Like, it’s not like you were in the bus lane, you were not supposed to be in this space because the bus might come, you know, through the space and you’re blocking it. It’s like those people are caught in, in fact, blocking the bus, not, not being where they’re not supposed to be, but like really being on the way of the bus. So that’s kinda like, you know, the, the perfect scenario of like the, the, you know, infraction.
Um, you know, so, so when it comes to the success of a program, like could you tell me a little bit when you have that conversation? And I know, you know, you, you alluded to some of the factors, right? Like the bus frequency and like there’s things that you cannot just measure everything the same because, you know, the bus is every 30 minutes and, you know, it’s not as frequent. It’s, it’s just you’re not gonna see the same thing than, you know, in other places that have a very frequent bus. So there’s a, there’s a sense of, um, expectation. Like there’s a bus here all the time, so I cannot park here because, you know, there’s a bus coming, versus maybe businesses that have, you know, goods being delivered and things like that, a-and then they’re like, “Oh, the bus is every 30 minutes. Just do this very quickly and get out of the way.” So that’s different. But, but could you speak back on, on how do you set those expectations for the agency? How do they gauge success? Yeah, I think the, the first step is really defining, um, the, the extent of the problem.
And so, you know, I think the initial deployment of the systems really helps us get a sense for the size and scope of the challenge that the agency is facing on a given route, and then what the impact of those, um, uh, of that, tho- those parking violations is on the, the network or the route in question. Once we, we’ve kind of established that baseline, um, we can really start tracking and seeing how it is we’re impacting, um, the, the number of, um, events on that route. Um, we’ll look at, you know, is the bus getting faster? Um, we’ll look at things like, um, uh, is the on time arrivals, um, increasing? We’ll also compare, um, the, the average speed of buses on the routes we’re enforcing to routes that aren’t being enforced. And, um, you know, sometimes even though it may look like buses aren’t getting faster, um, they’re not going as slow as buses on routes that aren’t enforced.
And so, you know, the, there is a, um, I think a, a, a way that we can, um, identify the benefits of our systems, while at the same time, um, recognizing that, um, we won’t always see all of the metrics we’re tracking heading in the right direction, um, because at the end of the day, these are networks and sometimes there are detours, sometimes there’s collisions, um, sometimes, um, there are changes to traffic patterns. And so over time, um, we just have to, to be able to settle on with our customers, um, some key metrics that we’ll use, uh, to understand, um, whether or not the, the program is bringing its desired, um, impact.
And, and speaking of customers, you all are, are growing. It, it seems like you’ve, you’ve got a lot more agencies that you’re working with, and as you mentioned earlier, five of the ten largest. It… That’s in the United States or
North America? In the United States. In the United States. So if, uh, i- if, uh, our research serves here, it, it’s over 2,000 vehicles that you’re currently tracking.
Is, is that… Does that still stand? A little more. Um, we’re really- More? Okay … approaching about 3,000 now, um- Okay … in the, in the US.
Um, and, um, that number will grow even more between now and the end of the year. Um, so, you know, we’re, we’re adding vehicles literally, or adding buses literally every single day.
Um, and, um, e- each, each bus we add, um, provides more data, um, that we can use to understand, um, how, uh, to more effectively, uh, deploy the systems we have.
A- and you mentioned buses, but my understanding is that there’s also trolleys that are involved too. Is that, is that correct, uh, with SEPTA?
Yeah, you’re, you’re absolutely correct. In, in Philadelphia in SEPTA, um, we have cameras installed on, uh, uh, uh, 38 trolleys.
Um, you know, tho- those trolleys are, um, I, I think it’s, uh, uh, uh, uh,
I guess unique in that while a bus can change lanes when it’s being blocked, um, a trolley really can’t. Um, a trolley’s on a track and it’s gotta stay on that track. And so when vehicles block trolleys, um, they really, um, uh, have the potential to, to shut down that trolley line for, or at least that segment of the trolley line, um, for a certain period of time.
And so the ability to, um, to capture those types of events, um, the ability to, to do so, um, in, in a way that allows, um, the agency to issue a violation without having to, to identify and dispatch a parking enforcement officer to a, a certain location, um, and hope that they, um, can get there in time to issue the violation, um, you know, it is really important.
And I think when it comes to, to driver behavior, um, you know, trolley blocking is a, a violation that, um, was rarely enforced prior to
Hayden AI, and we have high expectations, um, for how it w- w- you know, the, the deployment of our systems will help enhance and reduce, uh, the amount of, of trolley blocking that occurs on, um, Philadelphia streets.
Uh, it’s great. Uh, and you know, I can imagine that being a problem. I don’t… Not living in Philadelphia and only been there a couple times, I, I don’t know how prevalent that is, but I would imagine that it’s a pretty big deal and, and disruptive for folks that are using that system.
Um, you know, as you continue your deployments, and it sounds like the diversity in the p- deployments is growing, are there any, uh, any lessons or, uh, perhaps it’s- … uh, observations that the agency can gather from Hayden, uh, w- in
Hayden’s, uh, you know, in- intelligence, that, that camera system that maybe they didn’t know about before? It’s not necessarily blocking a lane or blocking a trolley or a bus stop. Uh, are you, are you hearing other things?
I mean, I, I think that, um, there are, there are m- multiple potential applications of the technology. Um, things that today, um, w- we may not even be looking at.
And so, I, I would say that literally every agency we work with, um, brings us new ideas and ways to utilize either the technology or the data that, that the, the technology, um, generates, um, to en- enhance their network. Um, you know, this is ultimately a vision-based AI system, and that system has the potential to see a very wide field of view and to capture data on, uh, everything from signs to, um, the condition of s- of bus shelters, uh, to, um, potential, um, obstacles that could impact a rider’s ability to board a bus. And we continue to work with our customers on non-enforcement applications of the technology that will ultimately, um, enhance the entire transit network, not just, um, identify vehicles blocking bus lanes or bus stops.
Yeah. That, that certainly makes sense, and I think that’s a, a really good, uh, good goal. Uh, you know, just kind of moving on from the US applications, uh, from what I’ve read, you all, you all also have some international, uh, deployments as well. Can you speak a little bit more about some of those projects and kind of what they’ve meant to those, uh, European cities that they’re in?
Yeah. Um, absolutely. You know, um, Europe, uh, is… While there’s no question that transit is an important part of mobility in the United States, um, it’s even, it’s an even bigger part of, of mobility, um, in, in Europe.
And, um, we’ve been fortunate enough, uh, to get to work with some, some large cities in
Europe on some pilot programs that, one, have helped us understand what’s different, um, in the, the European market, and moreover, what’s the same. And that knowledge, um, is, is helping us start to identify and figure out how best to deploy our systems in Europe and whether or not there’s anything differently we should do or know, um, based on the European experience.
Certainly, when you’re in the, the UK, um, you’re on the other side of the road. Um, so you know, that- that’s one, um, one thing we have to account for with our systems. How do we train our models to know that now you’re looking in a different direction, and things may look a little bit differently?
Um, in, um, you know, in, in places like Barcelona, um, there are a high number of motorcycles. How do we identify motorcycles, and w- how do we, we, um, capture, um, a plate on a motorcycle that may be in a, a, a- an area that is much smaller than we would expect, um, a license plate to be in? And so there are any number of model changes that, that have to be accounted for as you’re deploying in a, a new geography. Um, the localization of our systems, um, is something that we’ll continue to, to g- to, to build, um, and to grow.
But, um, working in Europe has given us the opportunity to, to test,
I think, our, our fundamental, um, belief, and that’s that the system is applicable, um, really anywhere.
And it’s proven that, that that’s true, that the same type of performance we see from our system, um, in the US, we see in Europe as well, and we’re very excited ab- the opportunity to work with cities, um, throughout, um, Europe. So I’m gonna ask you to pull out your crystal ball. Where do you think Hayden AI and this technology is in, let’s say, five years?
Well, I, I hope my, my crystal ball, um, is, is accurate, but I think that we’ll be on nearly every major transit network in the US and Europe.
I think that we’ll be providing not just, um, enforcement, uh, c- capabilities, but also, um, intelligence, um, that helps the agency, um, really manage their entire network. And I think we’ll be looking into, um, other types of vehicle applications, um, that are, are adding, uh, to the, the data lake, um, that helps give a more holistic picture about the types of, um, uh, of things that are occurring throughout a, an urban area, um, that impact transit performance.
Yeah, that, that’s certainly a, a, a great, um, application of AI, being able to, you know, use the cameras, detect what’s happening, and just help move those transit systems faster.
Uh, travel speed is, i- i- as well as frequency are some of the most important factors on a transit system. Uh, you know, i- if it’s not moving fast, not getting pl- people, places in a timely manner, then it’s not a great rider experience. So I, I think you’re doing phenomenal work out there helping move our different transit agencies across the United States, so, you know, big commendations for that.
Um, you know, as we’re coming close to, uh, the end of our podcast today, uh, we have a couple of segments, recurring segments that we do. The very first one is gonna be a rapid fire segment where we’re gonna ask, you know, short questions, quick answers, first thing that comes to mind. Are you ready for it?
I’m ready. All right. Favorite transit system? I’m gonna say New York MTA. All righty. Most overlooked cause of bus delay?
Road construction. Yeah. Those are very disruptive. Uh, bus lanes or transit signal priority? I think bus lanes. One metric every automated enforcement program should track.
Hmm. I’d say, um, repeat offenders. All right. Comebacks. They don’t learn their lesson. They come back for more.
Yeah. One application of vision AI in transit that is still underused. I think, uh, asset management or, or bus shelter condition. All right.
Uh, a city that has taught Hayden AI something unexpected. I think all of them have really taught us something. There isn’t one that, that stands out, although I, I would say, um,
I think London, uh, has taught us some things that, um, previously we didn’t, we didn’t think about. Could you elaborate, such as?
Uh, I… Well, I, um, certainly, um, driving on, uh, or capturing events on the other side of the, the road is, was new. Um, but I’d also, um, say the impact of, of roadworks on the network, and- Hmm … um, how significant that truly was, um, in a, a city like
London with, um, such a limited amount of space for the buses to travel. Interesting. Uh, and what’s one thing transit agencies should understand about AI before deploying it? Uh, I, I think that, um, the, the biggest thing they should understand is that AI can help them run their agencies more efficiently, um, but it, it still needs their input.
Um, that, that our system is very much, um, a system that brings intelligence, but also, um, needs to learn over time, and that learning comes from the experts at the agencies. Yeah. Well said. I really like that. Another recurring segment that we have, Charlie, is our, uh, key takeaways, and I wrote down a few while you were speaking. You let me know if I got any of these wrong or if you’d say them differently.
Maybe you have some that I d- I just didn’t write down, but y- you know, my absolute favorite was the three-legged stool concept. Enforcement, education, engineering, I think each of those have their place, and y- you know, the more that those are brought to the forefront from the agency perspective, you know, working alongside
Hayden AI, then you can start to change that driver behavior. So th- that’s a really important one, I think a highlight for me. Uh, a vehicle parked in a bus lane is much more impactful to the network than just your standard, you know, parking meter violation, right? I stayed too long in my spot. I’m,
I’m out of the way there, but if I’m parked in a bus lane, I’m very much in the way of the bus and the, the operator and the folks that are on, on board hoping to be able to make it to their job on time or to connect to another vehicle.
And another one that I’ve got is that, uh, traffic violations should trend downward over time. Once you get the, the education and enforcement sort of flywheel going, uh, people start learning, “Hey, I’m not… I shouldn’t do this. I shouldn’t park there. I’m gonna get a ticket.”
Um, or perhaps there’s an engineering, uh, application, right? Something that maybe divides a bus lane where you’re not able to park there anymore. Uh, all, all those things could, you know, continue to shape, uh, how people use the streets and hopefully keep our, our buses and trolleys, um, you know, going smoothly. And lastly, you s- you mentioned that there could be various applications that come out of some of these deployments that we just aren’t anticipating right now. Uh, things are v- happening very quickly, especially in, in the AI space.
Um, did I miss anything, or would you change anything of what I said? No, I, I think you, you nailed it. I mean, ultimately, on the, the last point, it’s about customer satisfaction, and we, you know, agencies want the experience for their customers, uh, to be as, as, as good as possible.
And so the time they spend waiting at the bus stop, it’s important that they know the condition of that stop. Is there broken glass? Is there trash? Are there obstructions?
Um, is it dark? Is there a light out? Um, those are all things that I think over time, um, a, a system like ours that’s visiting and passing these locations multiple times a day.
can provide significant insights, and, over time, that’s an area I think we’ll continue to explore. Excellent. Well, Charlie, this has been a, a wonderful conversation. I really learned a lot.
I, I knew some about Hayden coming into the conversation, uh, but I did not know, a- you know, all of the, the things that you’re, you’re doing, especially internationally, and, uh, you know, how the technology works. A lot of this is pretty new to me and, and very informative. I think it’s gonna be informative for our listeners as well. So really wanna thank you for spending the hour with us and, uh, you know, sharing your background and also talking about Hayden AI.
Well, thank you so much for having me. I really appreciated the conversation. Absolutely. It was our pleasure, and to our audience, thank you so much for tuning in this week. We’ll be back next Monday with another episode of Stop Requested.