SDS 256: Data Science in Transportation

Welcome back to the SuperDataScience podcast ladies and gentlemen, super excited to have you back here on the show. And congratulations on the anniversary episode 256 might not seem like a round number, but in the binary system, it’s actually two to the power of eight. So in the binary system that is, it’s written out as 100,000. So in the binary system, this is our hundred thousandth episode. And fun fact, the 256th day of each year is actually celebrated as the day of the programmer. And this year it falls on Friday, September 13th. So look out for that.

And today we continue our little saga of data science use cases in different industries. We’ve already talked about quite a few and you can check those out in the podcast archives, for this show. And today we are talking about data science in transportation. So let’s get straight into it.

Transportation is defined as a movement of humans, animals, and goods from one location to another via air, land, rail, or road, water, cable, pipeline and space. As you can imagine, it’s quite a massive industry in commencing a lots and lots of activities. The revenue of the transportation industry in 2017 was $5 trillion. That’s 5 trillion with a T, United States dollars, which represents about 6% of the total GDP of planet earth. How crazy is that? So massive industry, and let’s have a look at five use cases. Won’t be able to cover off everything, but we’ll look at five use cases and examples in each use case.

Use case of data science and AI in transportation number one, self-driving vehicles. So transportation is inherently connected to numerous other industries and we’ve already covered some of these examples, but nevertheless, self-driving vehicles are really the future of transportation. Vehicles on fixed routes such as trains and trams are operating more predictable environments and therefore there’s fewer challenges, whereas autonomous and semi-autonomous cars, ships and even airplanes are more difficult. But with the way, with the rate of technological progress, we’re getting there very soon. So here we’ve got an example in 2015, Boeing and Carnegie Mellon University launched the aerospace data analytics lab to develop a modern autopilot that will be able to respond to unpredictable situations.

Even though flying is statistically a very safe form of transportation, 55% of all plane accidents actually happened because of pilot error. So a better autopilot could make it even safer. And with the amount of data that’s coming in for a plane through the turbines to the wind speed, through all the sensors on board and, externally of, facing the external environment is a lot of data to process a huge, huge space for data science to innovate. And by the way, while we’re still on the use case number one, self-driving vehicles, I highly recommend to look out for episode number 259, which is coming up next week I believe, with Stephen. Oh, not next week, the week after that. So the week after that, episode number 259 with Stephen Welch and there you will learn a ton about self-driving cars. So I had an amazing chat with Stephen, can’t wait for you to check out that episode and make sure to look out for, it is episode number 259 all right, moving on.

Use case number two, traffic management. So the total number of vehicles in the world actually reached 1 billion in 2010. That’s crazy, 1 billion vehicles and that was nine years ago. Currently the number of vehicles is estimated to be about at about 1.2 billion and it is very likely to rise to 2 billion and beyond by 2035. So as you can probably gauge from your experience in your city, roads don’t double in quantity or volume ever simply because there’s not enough space for that many roads. Whereas the number of vehicles from 2010 to 2035 it’s actually going to double or thereabouts. And that obviously poses certain challenges for the infrastructure of cities.

So managing traffic well actually has never been more important or more difficult. So the problem actually is that one accident on the roads can cause a massive delay for lots and lots of cars and it can basically cause a traffic jam and in turn a traffic jam makes people more nervous. And it’s not a situation where it’s very obvious what to do. Sometimes you need to maneuver, sometimes you need to make decisions on the spot and therefore it can cause even more accidents and therefore it’s a very, very important problem. In fact, here’s our example. A data scientist at Nanyang Technological University in Singapore have developed a routing algorithm that tries to minimize the occurrence of spontaneous traffic jams. They are working with data provided by BMW car sharing fleet and claim that optimizing the driving patterns of just 10% of all drivers could significantly affect the entire network. So that’s where that cumulative effect comes in. Even if the driving patterns of just 10% of drivers are optimized, then the number of traffic jams can be reduced significantly and therefore the number of road accidents. Yeah, so another very important area in terms of transportation where data science could contribute and that’s traffic management.

All right. Example or use case number three, urban air transport. So we’ve probably all heard claims by companies such as Uber that we might have flying cars someday. And really that’s one of the fastest and easiest, probably not easiest, but one of the most obvious solutions to traffic jams and congestion and delays. It’s the fact that we’re operating on the surface of the world. So we’re operating, we’re moving about in a two dimensional space, whereas we live in a three dimensional space. So why don’t we go up and down as well. And of course that would require a lot of AI, data science, analytics and so on. And because the idea is here that with enough help from AI flying could become like driving a car. I know it might sound crazy right now, but if you think about it with the help of AI, it might not still not be simple, but it could be made like the most difficult parts of flying a vehicle could be taken on board by the AI and the human would just make the simple decisions of where to go and things like that.

Therefore, this whole process could be facilitated like in many examples of technology where AI or just technology itself helps us out and we operate on a more intuitive level. An example here is that in 2017, Uber signed a contract with NASA to develop software for managing flying taxis. They will begin testing electric taxis over Los Angeles in 2020 aiming to reduce the 18 minutes rush hour ride down to just four minutes. So look out for that. If you live in LA, 2020 is the year pencilled in for that. That’s very soon, actually next year.

All right. Example or use case number four, smart maps. So the days of paper maps are long gone. I really miss those days actually. Sometimes I want to go traveling, I intentionally don’t use my phone, I get a paper map and I try to navigate using it. Like we did with my family when I was a child and it brings up a lot of memories. It’s actually a lot of fun because how, how you get lost, how often and how much I get lost when using a paper map just cause I’m not used to it. But it’s a really cool fun activity to pursue I guess.

But going back to technology, we don’t use paper maps anymore. We use GPS assisted maps and in fact, it’s even hard to imagine a life without these maps. Sometimes when I walk, I use a map to find my way around city. On many, many times when visiting a new city, of course, I use Google maps and GPS for that, but that’s not the end of it. We’re actually getting to a stage where technology is not just telling us how to get somewhere, but where we want to go based on our preferences.

For instance, I have Foursquare installed on my phone and based on the past places I’ve been to for lunch and dinner, pasta, cafes I’ve been to, it can determine what I like, what my preference are. So I might be walking in a city and I’ll get a popup telling me where to go, what time might be interesting in checking out and of course how to get there. And that is actually going even further.

The example here is that Google maps, their mapping tool will soon begin using computer vision technology and your phone’s camera to provide visual directions when following a route. So basically, what you can expect from Google maps soon is augmented reality. So you’ll be looking at on your phone and you’ll be looking at kind of like the image of the road in front of you, like the camera feed on your phone, but it actually have augmented reality like arrows and things where to go and things like that. I think that would be pretty cool and that will really enhance the whole mapping experience and should be very fun. So again, that’s a lot of computer vision, data science, AI in there.

And use case number five, facial scanning. Surprisingly enough or not surprisingly enough in the space of transportation, facial scanning. And what’s this all about? Well, one of the biggest causes of delays and frustrations at airports is the security process, which is absolutely necessary, which is absolutely important. But if you’ve been to an airport recently, you will remember that it’s actually quite a lengthy process. That’s why they asked you to get to the airport about three hours before the flight if you’re flying internationally. And with the world becoming more and more interlinked with more planes and routes being developed, this is something that would be really nice if we could get through the airport security process faster and for it still to be safe.

Well, the example here is that a Swiss company, SITA or S I T A, installed their first automatic facial scanners at several US airports in 2018 and are working with US customs and Border Protection to bring the scanning process down to just three to five seconds, which will naturally speed up this whole security screening process at airports. At the moment, around 4% of passengers are wrongly rejected and further optimization will be necessary. But the start is promising. There we go. That’s another application of computer vision and data science in the space of transportation.

So there you have it, ladies and gentlemen, that’s the five use cases of data science and AI in transportation. Of course, those are just some use cases and some examples, there is plenty, plenty more. But if you’re in the space of transportation, hopefully this gave you some interesting ideas and insights. And if you’re not in transportation, maybe this will spark some ideas for you in your industry. And, I think there were some very interesting stats along the way. So on that note, you can get the show notes for this episode at www.superdatascience.com/256. That’s www.superdatascience.com/256.

If you know anybody who’s in the space of transportation or who’s interested in data science in this space, forward them this episode, this link and they can get these insights there too. And this brings us to the end of the episode. Thanks so much for being here today. I look forward to seeing you back here next time. Until then, happy analyzing.