Name: How A.I. can help improve traffic
Uploaded: 2022-09-26
Duration: 6 min 13 s
Description: A.I. is helping cities deal with traffic, providing crucial data on how to get the most people where they need to go in the most efficient manner.

A.I. is helping cities deal with traffic, providing crucial data on how to get the most people where they need to go in the most efficient manner.

How A.I. can help improve traffic

Tech

How A.I is helping solve the traffic problem

well. I haven't met anyone yet that really loves traffic. Ai is kind of a great technology to use because you can look at travel patterns and you can look at transit pad. My name is Tim Menard and I'm the Ceo and founder of light. I've been working on what's called intelligent transportation systems for over 13 years now, I've been on the automotive side of this for a long time using vehicle data to try and figure out how to solve traffic and connect others so that we can all get to where we're going safely. Many of our cities grew up organically and so they followed the streets and the street network followed some of the natural contours of the land and so a city like Chicago, which is in a very flat place. It's very easy to lay out a city that has a very regular grid, but in cities like Seattle or Pittsburgh, the topography means that the roads can only be arranged in a certain way. American cities are designed so that almost everybody has to drive like that's hard dependency. And so you design a city where everybody has to drive to do everything all the time. You shouldn't be surprised that there's a lot of traffic the way that we planned our traffic system for the last 50 years or so. The models that we look at for, for how we judge performance in the system, all kinds of other things are all based, how we're moving vehicles through an intersection, not how we're moving people through an intersection. There are professionals whose only job it is to move as many cars as possible and that's what they do. They just try to move as many cars as possible. But here in the Netherlands, they take a different approach. They don't try to move as many cars as possible, they try to move as many people as possible. And that seems like a very minor distinction, but it isn't the best example of how machine learning really gives us a great edge to optimize traffic is actually when we're doing it for busses and light rail vehicles. One of the things that we do often in North America is we'll have a bus right? But if you ask anybody, why don't you take the bus? They'll say, well, you know, I can drive to where I'm going in 20 minutes or it'll take me an hour and a half on the bus. If the bus is going to get stuck in the same traffic as your car, then who would ever take the bus? The number one thing that we can do is give transit its own lane at a per person basis. A dedicated transit lane can move 10 times as many people as an automobile lane and by giving it that space, transit becomes that much more attractive because now it can move in a very smooth way, it can really provide a travel time saving if giving it its own lane is not possible because maybe there's not a lane to give Providing some signal advantages at the intersections so that the signal knows that this is a bus that is carrying 30 or 40 people in it and that we should prioritize that bus more than we're prioritizing the cars that are carrying one or two people in them. So if the signals are smart, then you can move the transit a little bit faster as well. A traffic light uh, in North America is incredibly simplistic thing, fundamentally. Everybody gets green in one direction for a while, then it turns red and then everybody gets green in the other direction for a while and and that's it. But here in the Netherlands, traffic lights are more complicated. If a transit vehicle saying a tram or a bus is coming to the traffic light, they're going to get priority. If you're in a built up area, then the next priority will very likely be for walking or cycling. So if there's, you know, cyclists approaching the traffic light and the traffic light sees that there's a gap in traffic, it will turn the light green for those cyclists and let them through. Transit authorities are really interested. And also the pedestrian experience, because of if you're taking public transit, you're also walking those to actually go together. So your transit agency is actually looking at how do we make bikes, pet and your transit travel much more enjoyable, safer and effective and efficient. Imagine a system where you can quickly identify that there's been a minor crash that occurred on this street rather than continue to have a long string of traffic trying to scoot around the scene of that crash that you can quickly adapt your system to create an alternative route. Our whole platform is about taking vehicle data and connections to traffic lights and being able to make those green lights appear at the right time. That's gonna clear more cars out of the way so that you don't have that situation that we've all been a part of where a fire truck shows up light screaming, everyone's kind of panicking because everyone's stuck and doesn't know what to do. What we're doing is taking a bird's eye view and are getting all the lights set up ahead of time so that they get this nice clear path. So we drop a lot of things that make all the vulnerable road users much safer to in machine learning, there's still a lot for the machines to learn. You know, one of the big things that we need to tackle, our communication networks, all of this data. All of this intelligence consumes an enormous amount of bandwidth. We have so much data that's rich and available now that we can stop this philosophy of how do we react to something and we can start prescribing something and actually treating a traffic problem