Loading

Module 1: Intelligent Transportation Systems

Notes
Study Reminders
Support
Text Version

ITS Components, Applications and Communication

Set your study reminders

We will email you at these times to remind you to study.
  • Monday

    -

    7am

    +

    Tuesday

    -

    7am

    +

    Wednesday

    -

    7am

    +

    Thursday

    -

    7am

    +

    Friday

    -

    7am

    +

    Saturday

    -

    7am

    +

    Sunday

    -

    7am

    +

In this lecture, we are going to now introduce you to, uh, some of the basic taxonomy in its, uh, break down the ideas into its different components. Uh, tell you about how regular communication happens and also give you an understanding of, uh, big data. Use it in transportation. Uh, when we are talking about, uh, its uh, we usually are talking about at three different, uh, at three different levels. Right. We are either talking at the level of the vehicle or the level of the interest auditor cooperative level. So when we're talking at a regular level, we are talking about technology that can be deployed within the vehicles, including sensors, uh, processes as well as displayers, uh, with provide information to the driver. Right? So there are various, uh, uh, cameras that can be fitted within, uh, within, uh, within a vehicle. There could be sensors that allows the vehicle to not sweat too much on the, in the lane left or right. Uh, there could be, uh, parking sensors, right? Many of the vehicles now you would've seen, uh, uh, has, uh, when you put a vehicle in reverse gear, uh, it has a timer associate at the back of the vehicle that allows you to. See, uh, how much space is there and you can then back into, back into a parking space. So there are different all of those systems that are in the vehicle, uh, are the, uh, vehicle level, uh, classification of its devices. Uh, next is the infrastructure level, which are all the sensors that are on and by the side of the road that collect. Uh, information about traffic, right? So it can be your signals. It can be at the toll plazas that, uh, where your, uh, transactions are happening. Uh, so all of those kinds of devices that are either on the side of the road or, uh, on top of the road, uh, collecting information about. Uh, traffic are known as infrastructure, infrastructure level its devices. And then there are cooperative level, which is the communication between the vehicles and between the infrastructure and vehicles. Right? So then there has to be also, uh, it is one thing to have just a vehicle level or an infrastructure level, its uh, capabilities. But unless, and until there is communication between the. Uh, vehicle and the infrastructure, uh, your efficiency is most likely not going to be achieved or higher, higher efficiency is not going to be achieved. So there has to be certain two way communication between your infrastructure and your vehicle. And so, uh, for example, in the last lecture we have told you that how two way communication can be achieved of an ambulance or an emergency vehicle at a signalized intersection. Right? So the signal detects, uh, the vehicle that is. And for, uh, that is, uh, uh, communication from the, uh, infrastructure to the vehicle. And then the vehicle communicates back saying, I am an emergency vehicle, give me a right away. So then the, uh, that is communication from the vehicle to the infrastructure. And then the infrastructure, uh, turns on the signal turns green. So that is kind of two way communication. So that kind of, uh, communication is done. Cooperative level ideas devices. Uh, when we look at the, uh, classification based on three phases, uh, what we see is that our users on one side, and then there are all the strategies and the planning, uh, segment on the other side, right? So, uh, users want to use users want to, uh, have an efficient transportation system. Uh, government agencies want to. Uh, device that policies were developed, their priorities. Uh, based on, uh, the existing system and, uh, in between are all the different its architecture that needs to be developed in order to satisfy both ends both the users, as well as the, uh, planning agencies or the government agencies. Right. So, uh, when you look at night yes. And we'll give, we'll give you an understanding of what is a logical architecture and what is a physical architecture. Uh, these are, these are basically how you are. Uh, information flow happens. Uh, what are all the, uh, hardwares and the softwares that are needed in order for a system to be created, right? Uh, should the information be only one way flow, should we should be only two way flow? What kind of information should be shared between, uh, different, uh, uh, between the different, the stakeholders so on and so forth? So all of this is a very interesting new field, a new arena in which, uh, into which transportation. Uh, professionals are getting into, like I said, in the previous lecture as well, and they should, uh, be, uh, something that, uh, opens up new avenues, uh, for you as well. So for example, uh, this is, uh, we have already told you about it. What a traffic management center is. The one in the previous lecture that we had showed in Bhopal is the first traffic management center that has opened as a part of the smart city. So, if you were to understand, or if you were to understand how information flows between the traffic management center at various other entities, you would have to develop something called a physical architecture, uh, which shows you, which shows the traffic management center and how it is linked to the roadway. Right. What information is it? Uh, is the traffic management center sharing with the roadway. And what is the roadway sharing back with the traffic management center? Furthermore, how is the roadway then connected to the driver as well as to them vehicle that the driver is in? So what information is the roadway getting from the, or giving to the vehicle and giving to the driver? So is it getting back anything? It doesn't seem to be getting any information back from these. So it is there one way type of information that is being given similarly roadway, network condition information is going out from the traffic management center. To the different ISP, right? So Google a dollar, getting all the information about the roadway network condition from these TMCs. And it is a one way. Information that is going out. Similarly, there is a lot of, uh, uh, one way information that is maybe coming into the TMC is the maintenance and construction manager. So suddenly there is that a roadway construction that is going, uh, in one part of the city. That information must be only coming in to the TMC, which is done. Maybe. And working that TMC, uh, that information into some type of travel time or travel delay, uh, and then forwarding it to the, uh, information service provider. So you see, in order to S this is a very good diagram that, uh, is a very simplistic diagram that allows you to understand how a, uh, how an it system. May work. So this is a, this is an idea system, meaning this isn't traffic management system, right? So this is a traffic management system. How is it connected to the roadway? How is it connected to the basic drivers? How was it connected to the, uh, for example media as well? So if, uh, if you have to, uh, provide information on the TV or on the radio about the traffic conditions, uh, it can, uh, they can direct to get information from the traffic management center. Uh, uh, traffic operational personnel can get information from. Uh, from this, and of course, uh, different, uh, uh, bus transit managements, and all that emergency management can get information on this. So the basic thing to understand is who are all the stakeholders that need to get, uh, get data or, uh, from whom we need to, uh, receive data, what kind of data we need to get and does the data have to be a two way communication or one way communication? And so on and so forth, these basic things, if you, if one understands, then one would understand how to develop its system for their city. Uh, what are the basic competence, uh, of its efficiency, right? How does it improve the efficiency? Uh, in, uh, in your urban traffic network, first is it allows you for automated data acquisitions, right? We've already given you a brief understanding of that. Uh, automatic vehicle identifiers are now nowadays used, uh, especially in a public transportation, uh, arena where you can, uh, add, or you can have these Avi sensors on your buses and you can easily locate. Where the bus is, uh, based on that you can identify, uh, you can put up information at various bus stops as to when the next bus is arriving and so on and so forth. Uh, you can also, uh, use GPS technology, automated vehicle locators, AVL, uh, which is, uh, uh, is, uh, provides fast, flexible, and, uh, relatively inexpensive data to determine a vehicles position. Well, as it in real time, right? You can use that, uh, fast data communication to the TMCs. Now, once you have that, that data has to be now communicated to the TMCs as well. So different types of communications, uh, protocols can be used, uh, DSRC dedicated, short range communications, uh, come continuous. Yeah, the interface long and medium range and media to public like VMs sharp as in science, right? Uh, VMs is nothing but variable message signs that you might've seen on the side of the road saying that, uh, slow down, uh, traffic ahead or slow down a accident ahead or something like that. That can be easily controlled. Those messages can be controlled. From the traffic management center, based on the data that it is receiving from, uh, vehicles are, uh, yeah, incidents downstream, uh, the other way, the other way, the components of its efficiencies, accurate analysis of the data. Now, when we're talking about all this data, remember this is, we're talking about big data, big data, meaning large volumes of data. Now from these large volumes of data, you have to make sense of what data, how much data you need. And only based on that data, that much volume of data you have to convert now the data into information, right? So it is only, it has no, uh, it is, uh, not very efficient to just have CCTV cameras everywhere. Right? You have to convert the feed of the CCTV cameras into some sort of information. Uh, so if you have, uh, if you, uh, if you have CCTV cameras and you are looking at the, uh, at the intersection, Uh, but you are still manually, uh, changing the signal timings, uh, that does not help you, uh, make improve the efficiency of that, uh, traffic signal. For example, maybe all that, uh, information, uh, that, uh, for traffic signal feed that you're getting, you have to then try to automate, uh, as to when, how to change the traffic signal automatically based on. Data that is coming. So that is an example of, uh, what we mean by saying accurate analysis of the data. Right? So poor data has to be weeded out. Data has to be cleaned. There'll be a, not a lot of noise in the data, some data outliers. So we have to check for the outliers as well, whether they are useful information or not. All that has to be done before to just establish whether your, uh, uh, system is able to capture the current state accurately or not. Once it is able to, once it is able to capture the current state, for example, or in other times, once it is calibrated only, then you can use it for forecasting, uh, traffic for different States. Uh, different traffic conditions and so on and so forth. Uh, next it has to provide reliable information to the, uh, uh, public as well as the traveler. Right? So just providing information is not good enough. It has to provide a reliable information. For example, the next bus is arriving in, uh, arriving at 5:15 PM. Now, if that 5:15 PM has to be reliable, if you don't provide reliable information, What usually tends to happen is, uh, the user will no longer have faith in what you are telling. And, uh, once, uh, that fate is lost, then the user may not be, uh, may not use your services. Uh, that is one loss or the bigger loss. Maybe that the user may not use public transportation system. Uh, on a whole, uh, which is a bigger loss now that then he or she may be shifting back to the private vehicle. Right? So reliability of information is very, very crucial, uh, that improves the efficiency of that allows, uh, its to improve efficiency of our transportation network. Now, what are the different types of vehicle communications? We're already told you a little bit about it. Uh, vehicle to vehicle communications. Uh, right. So this vehicle is communicating to this vehicle saying that I am in seconds behind you, for example, right. I am N seconds behind you. So maybe it's a time headway. What we call it this time headway. So I am in seconds behind you. So that is vehicle to vehicle information, uh, vehicle to infrastructure communication between vehicles to the infrastructure. So for example, uh, this vehicle may be communicating to. Uh, uh, let me see. This vehicle may be communicating to, uh, awake, uh, uh, a, um, a base station or a mechanic, um, audit auto to the auto, the PMC, a traffic management center saying that I am broken down. Right. Uh, my vehicle has broken down and this is my location based on, uh, cell phone towers. Uh, and I have to be, uh, I need some assistance at this location, right? So this is communication from the vehicle to the infrastructure, to some infrastructure for in this case. Uh, to the TMC or to somewhere else. Right. And then we can do everything that means vehicle to any entity that, that it can influence it. So it can send vehicle to vehicle, to vehicle record infrastructure vehicle to, uh, Uh, pedestrian who was walking next to, uh, the vehicles or to anybody who can influence the vehicle. Right. So there has to be, that is the ultimate, uh, aim as to, uh, the vehicles should be able to, uh, communicate to anybody who can influence it. So those can, that is still, uh, we are still, uh, at a nascent stage in that, uh, however, there's a lot of vehicle to infrastructure communication that is happening, uh, vehicle to vehicle communication. Some, uh, vehicle to vehicle communication is, is happening and we'll tell you how it is happening. So when we talk about vehicle to vehicle, communication is usually happening through, uh, wireless networks, uh, WPN networks, right? Allowing vehicles in the, uh, in transit to transfer data on their position and their speed, right. Position and speed. These are the two, uh, data points that, uh, vehicles communicate. Uh, what happens is that. Uh, for example, a driver of a vehicle can receive a warning in the event of an accident risk, or the vehicle itself can independently take preventive actions. So you may see that, uh, modern cars are coming with automated brake baking system. Right? Abs so what it says it says is that if the, if your vehicle gets too close, Uh, to the vehicle in front of you, or if the vehicle in front of you suddenly breaks and he's getting too close to you, uh, your vehicle automatically asks you even, you don't have to put your foot on the brake, but the vehicle automatically breaks, uh, are, uh, there are, uh, onboard sensors, cameras and radars that allows you to tell, uh, that that tells you that a vehicle is too close. Uh, on your right hand side. So don't change your lane right now, right? Um, you know, maybe your vehicle is in your blind spot, so do not, uh, change your lane right now. So these kinds of sensors are already in use in many of the vehicles. Uh, which allows you to, uh, have a safer driving experience, right? So they improve your, the safety of your, uh, of, uh, of the, each of the passengers for each of the drivers and in turn. So when safety is improved in turn, the efficiency of the entire network improves, uh, vehicle to vehicle communication. The other thing is autonomous security systems are like, we've already said a blind spot monitoring. Uh, automatic braking systems, uh, automatic emergency brakes, lane departure warning systems. So we have already, uh, discussed about these three, these three things. And these are already some things that are available, uh, at the high-end, uh, in some of the high end models of the existing, uh, vehicles that are even available, uh, in India as well. Now, vehicle to infrastructure communication V two, I, uh, components include RFID readers, traffic lights, cameras. Uh, lane markets, streetlamps, signages, and parking meters. Right. Uh, so, uh, parking is something interesting that is happening in some of the urban areas, uh, as well. Uh, you can have on-street parking that can be, uh, monitored, uh, using, uh, uh, information using vehicle to infrastructure communication. So, uh, uh, on-street parking. Uh, lanes are created on street parking, uh, um, zones are created where, when a vehicle enters into that zone, uh, it is automatically sensed. And then, uh, there is a mechanism by which, uh, it gets, uh, a meter is, uh, uh, starts automatically for parking price. And then, uh, based on how long your park, you can automatically get billed. Uh, for your parking. So this is something that is, uh, being used, uh, in the case of, uh, some of the cities, uh, that are looking at, uh, uh, efficient parking management systems, uh, for the city, uh, wireless bi-directional and, uh, and similarly vehicle to vehicle using dedicated, uh, Shortland communication frequencies to transfer data. Uh, DSRC is something that is very, very necessary in order to transfer data. Uh, we do eye sensors can acquire infrastructural data and provide travelers with real time advice, right? Uh, vehicle to infrastructure, the most prominent ones are your traffic signal, um, uh, that we have already discussed about how the traffic signals can. Uh, no, I mean, uh, we can, uh, significantly reduce, uh, the presence of, uh, uh, traffic police at a signalized intersection. Uh, by using such, uh, ICT devices, uh, traffic supervision and management systems can be used, uh, can use the data collected from the infrastructure and records to set various able to speed limits and adjust signal phase and timing, right. To actually feel something. So, yeah. Uh, based on the arrival patterns that you see, uh, through your CCTV cameras at the TMCs, You can then set your traffic signals more efficiently. So that is what we have already mentioned to you about converting the data that you're getting, the feed that you're getting from your CCTV cameras into information, and then processing that information and using it to forecast. Uh, what your travel time or travel speed or travel delay is going to be on the corridor. So that is something that is very, very crucial and, uh, ICT can help improve, uh, the efficiency of, uh, our help people, the efficiency of a signalized intersection. Uh, vis-a-vis reduce the delays at such intersections. Uh, wake up to X vehicle to everybody, uh, vehicle to vehicle and vehicle to infrastructure. Communication models mentioned earlier. Uh, completed, uh, in the V two X, which represents a generalization, a data transfer from a vehicle to any entity that can influence it, or vice versa vehicle to pedestrian vehicle, to roadside vehicle, to device, through to grid. So this is something, uh, that is being still worked upon. Uh, uh, this can alert a passenger, um, uh, a pedestrian, for example, who is trying to. Cross a street at an unsafe analyzed interests action. Uh, and, uh, the vehicle, uh, can now communicate with, with the, um, uh, person who is trying to cross the street and maybe send a message comes to her, his or her mobile phone saying that vehicle is approaching. Uh, please be alert. Uh, this is especially, uh, situations because we at Rand signalized intersection, we usually. Uh, nowadays, uh, either, uh, have our headphones on, we are not looking at the traffic and in such situations, these things might be, uh, helpful, right? But these are still in a very testing, uh, phase. Uh, and this is something if it becomes, uh, operational or if it becomes a viable can help, uh, in the safety of, um, various safety of the transportation network as a whole, at various locations in our network. We would need to develop such warning systems for, okay. Now when we talk about big data, let us give you an understanding of what we usually are trying to say. Make big data. When we talk about big data, we are usually talking about a mobile or cellular phone data. Uh, there are three V's associated with it. One is volume like large dataset when it's velocity high speed. Of data acquisition, the speed at which the data is generated and acquired is very high and variety is a mixed data type. There's a lot of different types of data that is coming in. That's why we say, uh, big data. Uh, first and foremost that is needed in big data is, uh, data cleaning what you may be getting a higher, you may be, for example, you may be getting a hundred pieces of information, but maybe all you need is 20 pieces of information. Right. So you have to screen the data and we doubt those other 80, uh, pieces of information, and then only look at the 20 pieces that you need. But in order to do that, you are not only getting 800, for example, like I said, maybe you are getting one lakh per minute, right? And if you get, if you talk about only 24 hour data, you may have huge datasets. So you even just for working with. 24 hours data and clean up that data. It'll take you a lot of computing. Uh, it'll take you, uh, of computing resources and computing knowledge that you will need, uh, in your, uh, TMC or in your team. Yeah, it'll be very, very high. Right? So volume cylinder, phone data generated as a result of phones, communication. Uh, one thing, uh, it gives you is the positioning. Uh, after, uh, um, after the, uh, of the various cell phones, right. Uh, positioning, meaning, right. You know, all this, we have already talked about how cell phone towers and, uh, can be used, uh, in triangulation and trilateration to find out where, uh, an object is usually yourself from your cell phone data. Two types of data can be. Collected for, uh, transportation purposes, uh, call duration, the card CDR or citing data, right? Uh, we'll tell you what, what each of those things are, uh, uh, called, uh, CDR or call duration records. Usually it gives you a lat-long, uh, of your particular Mac ID. Uh, what time, uh, that Mark ID was located at that location. And for how long your call went on for, right. So, uh, in, through this information, what we can tell you, uh, what one could understand is that nobody can trace it back to you because see all the personal identification, uh, features have been anonymized. Uh, only we know there's a certain phone with a Mac ID of this, uh, that, uh, started a phone call at this time. And that phone call lasted for 81. Uh, seconds. So we can come to know about his or her, uh, call duration record sightings is whenever a cell phone is positioned, detected by a tower, right? This is just sightings. You don't have to make a call in order to find out where you are. The, uh, the cell phone towers can just detect. Uh, if you have your GPS signal on, on your phone, the cell phone towers can just detect. Uh, where you are, uh, based on again, triangulation. Right? So, uh, this ID, uh, was detected at this time at that location based on triangulation. So those are citing sightings are a little bit more refined than CDR and can do a lot of temporal and spatial resolution. Uh, CDR usually has to, you have to make a phone call in order for you to locate, uh, the person in this case, you don't have the, how make a phone call. You just have to have your GPS signal on in order to be able to track. Right? So, uh, for example, uh, what is then, so CDR and sightings data can be then converted into. Uh, for the information which, uh, the traffic, uh, uh, or transportation engineers are, uh, usually interested in is the origin and destination data, right. We are always interested in whether a person is. Uh, where a person is located and where he or she is traveling to. That is what our primary concern is. Uh, our primary concern is, uh, whether he or she is, uh, located in residential zone, located in a commercial zone, located in a retail zone. And how long does it take him or her to travel between different zones? That is what essentially we are trying to find out. So, uh, from the CDR data, you can then, uh, Find out different activity location in your city as well. Right? For example, if you are seeing that this ID J zero one, uh, is, uh, are making different calls during these times of the day, right. And from these same X, Y coordinate locations, right? So usually what you can. Infer is that these are night times, uh, evening nine, 9:00 PM, 10:00 PM, uh, 11:00 PM at night. Uh, midnight later, midnight. So usually what that would should tell you is that wherever this person w this location should be his or her home location, right. Usually a person, uh, should usually a person would not be in his, in his or her office such honor, such a such later time in their day. Usually he or she should be at home usually. Right. So this is how now you have, uh, this. But this IDs, uh, um, data for about a year or so, I guess, right. Once you have it for a year and then you can generalize. And see, yes, this may be his or her home location are, then you can have multiple of such ideas and then figure out where each of them, their home locations could be. Similarly, if you have data during the day times, and then you can figure out, uh, during the daytime JS, one is making different calls. From a different X, Y location, maybe that location during the daytime could be his work location. So similar. So you see now you have an information as a transportation engineer or traffic planner. Now you have information about your origin and destination of your different users, uh, based on just. Uh, cell phone, uh, uh, data, uh, calling data records, right? So this is something that is, uh, improving vastly the volume of, uh, uh, data that we are getting otherwise for two other, uh, in order to get origin destination information, we used to have, we used to conduct only, um, uh, uh, questionnaire surveys and the sample size would always, uh, I have to be would inevitably be always low, so not give very good information, but now, uh, through, uh, its devices or through cell phones, we are able to get larger volumes of data that can validate our claims suffer from our questionnaire survey that we weren't getting there can validate it. Of course. So, like I said, the activity location home can be inferred because the frequency is four out of seven records show that he or she's making these calls and these different times. So with certain amount of confidence, we can say that that is his or her home location. Now how can citing data be used? Uh, citing datas can be used, uh, to convert, uh, or to, uh, data and the activity locations by something called distance based clustering. If you have taken or you're encouraged to take more and more statistical classes, especially if you are, uh, if you want to work in its, because this involves a lot of big data analytics and statistics becomes more and more crucial. Uh, in your ability to be able to work with such a large volumes of data. Uh, so distance based clustering, one of the methods is, uh, k-means, uh, based, uh, distance based clustering. What it essentially tells you is that if there are, uh, key points, uh, with initial centroids, uh, certain centroids, and if these centroids don't change after, uh, after reforming, uh, Uh, newer clusters, then you can combine those combine, those points into one cluster location, right. There may be different points, but can you say that this is one cluster and this is another cluster? Or can you say that? Do you want to say that this entire thing is a cluster? How do you determine. What is a cluster. So this gaming's a distance based clustering is a very, uh, simple, uh, method of, uh, method by which you can determine different clusters. And once you know, different clusters, then you can say that this is a home cluster versus this is an office cluster versus versus this is a shopping cluster. And that gives you an understanding of, uh, origin and destination in your city. Right. So let us see an example, uh, that can allow you to, uh, simplistically, allow you to understand how, uh, k-means clustering works. So consider a sample of a weekday citing data. The geographical lat-long has been transformed into Curtis in X Y coordinates. So these are mostly X Y coordinates for simplicity. The time is also simplified from Unix to. Uh, etcetera. MMSS so this is our minutes and seconds use k-means clustering to group the scientists into two groups and determine suitable activity locations for the same. Right? So you have, uh, one ID, uh, and for that one ID, you have been tracking that one ID and seeing that. Uh, it has been at, uh, these different locations during these different times. So now you want to, based on this location data and time data, you want to see, uh, if there is any pattern, uh, if there's any activity patterns that you can determine, right? If you can determine that they are, uh, at a home location and an office location of what is happening and where, especially, uh, there could be, especially you already know that they are different. Uh, lat longs. So, uh, can you determine, uh, conclusively based on this information, that, which is the home location and which is the work location. So what essentially, what first you do is you assign initial values of center-right, right. Uh, just pick based on, based on what is available here, uh, of these different, uh, lat-long, uh, of these different Cartesian coordinates. Uh, I go X, Y plane in the X, Y plane, or you just have initial values since K's only two, you said only two groups. So let us, uh, assigned two centroids at the first center.