Q. I'm curious if you share any of this speed data with the law enforcement community or the state troopers there in Georgia.
A. Hugh Colton: There was a bill passed back in the 1990s to prevent the ITS equipment being used for law enforcement. It was led by the then speaker of the house.
Q. Will you be using public transportation ridership as part of your variables? Also, how would this product be used for a public transportation radio room to re-dispatch paratransit service?
A. Marcus Wittich: The first part of that I can talk to. Actually, the only regression analysis work that I have seen in my research to date relating to congestion and other variables was done with relation to public transportation ridership. So to answer that, I can't tell you exactly all of the different variables we'll take in account, but I think that's a good one to take in account-specifically looking at the fact that somebody has already done some of that work, trying to create that statistical analysis. What was the second part of the question?
Lokesh Hebbani: How would this product be used for a public transportation radio room to re-dispatch paratransit service?
Marcus Wittich: I don't have an answer for that one. One of the things we're doing right now is keeping a very narrow focus from our concept of operations to use it for one purpose, helping us stay on track. Not knowing much about paratransit, I couldn't answer that. Hugh, do you have any comments related to that?
Hugh Colton: Other than I have read papers which have talked about using some sort of computer system to do this, I really don't have any information I can pass on to you.
Q. How old are the field devices and what percentage are actually working correctly? As Hugh mentioned before in his presentation, the weekly performance report usually gives how much percentage of the devices are working. Could you elaborate on that?
A. Hugh Colton: Looking at the most recent data, unfortunately I don't have it with me, but I have a copy from a couple of months ago, and we had 82 percent of our CCTV cameras working, 77 percent of our changeable message signs working, and 100 percent of our ramp meters working.
Marcus Wittich: As far as the age, some of these date back as far as--
Hugh Colton: Yes, some of our changeable message signs and some of our cameras date back to when the system was installed and it was commissioned in about April 1996. Since then, we have replaced some of the changeable message signs. We have replaced quite a few of the CCTV cameras. And we are in the process now of letting a maintenance contract and one of the first elements of that maintenance contract is the renewal of pretty much everything that's more than a couple of years old. All our old style changeable message signs which used fiber optics are going to be changed out for LED technology. Our old, large pan, tilt and zoom cameras are being replaced with dome cameras. In the last two years we have almost doubled the size of our coverage, so almost half of our equipment is no more than two years old. So we don't have a great deal of old equipment except for changeable message signs, which are quite expensive to replace.
Q. What performance measure do you intend to use?
A. Marcus Wittich: That's a pretty general question. From an outcome point of view, we're looking at performance measures with relation to congestion, to incidents, and to public perception. The congestion performance measure primarily will be a travel time index, but we'll also be putting out travel times. We may be putting out some buffer time indices. From an accident point of view, we'll certainly be doing number of accidents per segment. Again, we're in the middle of this, so for a lot of these we're starting an iterative process. We haven't narrowed it down. But a lot of the idea is to have metrics that can be comparable across segments. So number of accidents per segment isn't comparable, because segments have different lengths. But we're looking at a number of different metrics there. As far as output metrics, those follow once we've defined our outcomes. Those are still to be determined and coming. We have lots of ideas and we have our current ones we're using. But unless they can be linked to the outcomes we're interested in, we're going to avoid using them.
Hugh Colton: Of course, this contrasts quite drastically with our current performance measures, which are really measuring how many calls our operators take, how many incidents we responded to when I guess the smaller the number, the better you're doing. Because if people aren't calling up and asking for help, then the system is working well. We know that we've got to handle peaks when there are many calls, but our aim is to reduce the number of calls coming into us and reduce the number of incidents by improving the safety.
Q. Did I understand correctly that you suggested starting the work on your process before seeking upper management endorsement? Isn't that backwards?
A. Marcus Wittich: Here in Georgia, the operations of the traffic management center is outsourced to Serco. We are tasked to running those operations floor. In that, we have leeway to put our own processes in place to improve it. In essence, I am upper management for that process.
Q. How much behind the scene data processing is involved in converting the raw data to the performance measures output?
A. Marcus Wittich: That's a good question. This is one of our challenging areas. We have many years of data, lots and lots of congestion data. It's all stored on flat files that are not accessible. The last four to four and a half months of struggle has been working with GDOT IT to write a pipe to extract that data and then there's a bunch of work to normalize that data, to smooth it out or to make corrections for when sensors are giving bad data and there are gaps in the data. We're actually getting very close to the end of that first stage. But it's been a considerable challenge and effort. I fully expect as we continue to add new performance measures and gather data for it, it will be just as challenging.
Q. The question is: What is the anticipated timeline for implementation of these performance measures?
A. Marcus Wittich: We're nearing the release of our first iteration of this dashboard now. I had originally targeted that I wanted to get that done by the first quarter of this year. As with many projects, this slipped, one, due to the funding, and two, due to some dependencies. I fully expect the very first iteration, which is only going to be output and congestion data, but has a lot of the underlying types and methodologies to get data and the initial user interface should be out in the next month or two. I expect to follow that every two to three months by another iteration where we're going to add another performance measure to it. The entire process actually doesn't have an end in sight, but I expect it will be several years before we get a fully mature, rounded release. It's a continual process. It's there to support continual process improvement and I don't expect, as of any good piece of software, the lifecycle is continuous.
Hugh Colton: But we'll see some results later this year.
Marcus Wittich: Yes. Our first iterations will be coming out within the next month or two.
Q. How long has your statewide ATMS been in existence?
A. Hugh Colton: The system has been in existence since April 1996.
Lokesh Hebbani: During the Olympics?
Hugh Colton: Yes. It was implemented prior to the 1996 Olympics.
Comment from remote audience: I commend you on the framework that you have developed consisting of solid foundation and pillars. It would be nice to have a roof, too. Here is how I see complete performance management, hierarchy of mission. The most important thing in organization for a system is trying to achieve. Example: Georgia DOT provides a safe, seamless, and sustainable transportation system that supports Georgia's economy and is sensitive to its citizens and the environment. Goals: What does an organization specifically wish to accomplish during a time period, such as a year, three years, or 30 years? Example: Support Georgia's economic growth and competitiveness. The T3 objectives specific, quantifiable steps towards achieving a goal. For example, move Georgia citizens towards school and play reliably, affordably, and within an acceptable timeframe. One other performance measures here: quantified outputs and outcomes supporting specific objectives and demonstrating progress. Here, I think they have given us some information about what can be done, which you have explained through the presentation, but I think they have given a very good explanation of what else can be done.
Q. Your performance metrics are freeway-centric and driven toward freeway incidents, level of congestion, and flow dynamics?
A. Marcus Wittich: Is there a question there?
Lokesh Hebbani: Your performance metrics are freeway-centric-They want to know whether it's freeway level of congestion or flow dynamics.
Marcus Wittich: The performance measures in the three categories we chose, which are congestions, accidents, and customer satisfaction, are all targeted at the freeways because these performance measures are related to support improvement in the traffic management center operations, which its goal is, and focus is on the freeways.
Q. How do you communicate to the various statewide ITS devices? Either by fiber or wireless or through modems?
A. Hugh Colton: The vast majority of our equipment, which is in the Atlanta area, is connected through fiber optics, either discreet fiber optic connections or, more recently, through Ethernet connections. We do have a few dial-up devices where we don't have fiber. There are a very small number of wireless devices, typically hopping across to a nearby fiber connection.
Q. What are the typical methods used to acquire public perception data and how is customer satisfaction measured in your system?
A. Marcus Wittich: As I mentioned, that's one of the challenging areas we haven't solved yet. Of the thinking that we've been bouncing around, one is that we actually house a 24/7 center where we take calls from the public. One of the thoughts was to actually poll them, ask them questions, or direct them towards a website. Some of the other measures that we have gotten are longer in term. We have actually done some studies where we've actually asked the public. Our need for this is more real time. By the way, those studies we've done were surveys. Our target to make this work has to be more real time. So either we need to direct people towards a website, possibly when they call in 511, put the 511 where they can actually have a satisfaction hit button 1, 2, or 3, like it, don't like it. Again, that's a challenge we fully recognize is going to be difficult. We'd like to set our goals high.
Q. I know other DOTs, such as Washington DOT and Florida DOT, are also working on this data collection. Are you looking to see what they are doing?
A. Marcus Wittich: Absolutely. We have been scouring the country for best practices. One of the big gaps that I've seen, and our vision is, we really wanted to isolate the impact of what we do on our outcome metrics and through some sort of regression analysis. We have not been able to find anybody in the country that's doing that. We did find one study doing some regression analysis with relation to public transit and congestion. But yes, beg, steal, and borrow from anybody. We have no desire to reinvent the wheel, if it's already been invented.
Q. What metrics do you plan to use to measure highway network performance? Also, how do you plan to display the results of the highway network performance so that it is meaningful to the user?
A. Marcus Wittich: First, to answer that, the user in our initial target is our traffic management center operators. We're not targeting this to the public. We know, in the future, very likely this may morph to that, but that's part of the beauty of our concept of operations, to keep us narrowly focused. I believe we actually covered the first part of that question at a high level. What are the metrics we're talking about? Possibly travel time index and a number of others, but it's an iterative process. We don't have them all mapped out—we'll be developing them as we go.
Q. We have heard general concepts. Where are you with respect to hard design aspects, that is, real measurements and the correlation of those to your outcomes? What are your precise average travel times, average speeds on the geographic roadway segments? Of course, these are really only controllable with ramp metering and some of the DMS for load spreading. I would be interested in knowing what hard measures you plan to use for this program.
A. Hugh Colton: I guess one of the most important things we do for keeping traffic flow is our HERO program. This is a roadside assistance program which has the foremost object of getting the travel lanes open when there are incidents. We have, in fact, reduced the time for incident clearance by two-thirds. We also have another program which is called TRIP. The purpose of TRIP is the towing and recovery incentive program. Traditionally, when you had a major wreck involving commercial vehicles, you call in a wrecker company and maybe also a cleanup company and they get paid by the hour. So the longer they can drag out the process, the more they're going to invoice back to the company that had the problem. We've turned that on its head. What we've done is we've got three trained wrecker companies which have had to meet certain national standards. When there is an incident which either a police officer or a HERO supervisor or a GDOT foreman decides warrants a TRIP, we call TRIP and the clock starts ticking. We pay a bonus to the clearance company on the basis of their performance. If they clear the road from the time they get the message to start clearance—and that may be some time after they arrive on the site, because maybe fatalities or having to deal with fuel spills and so on-but they then have 45 minutes to get the road clear. We have been running this now for 17 months and only two TRIPS have we failed to pay out. The bonus is up to $3,600. It's a real incentive to get on and do it. We have five TRIP-certified companies at the moment and the environmental cleanup company. It's really a pleasure to see how well they work and get the job done and get the roads open. I'm getting some numbers off the top of my head here, but I think the average clearance time was up around 261 minutes. It's now down to--
Marcus Wittich: 59 minutes.
Hugh Colton: 59 minutes. So we've gone from 261 minutes for a clearance to 59 minutes for a clearance.
Lokesh Hebbani: Also, I think we started this program on January 2008.
Hugh Colton: Yes, so it's been running about 17 months.
Lokesh Hebbani: Yes. Since then we've had about 84 activations.
Hugh Colton: Right.
Q. Do ramp meters really improve access to the interstates? Please explain how.
A. Marcus Wittich: I would probably first say that's a topic for another webinar.
Hugh Colton: You have to understand what the purpose is of a ramp meter. A ramp meter is not a silver bullet that's going to stop congestion. A ramp meter is going to delay the onset of congestion. So if you are running over capacity, you are going to have congestion. But I think we have good evidence that we have been able to-- if you draw that graph, which is usually the big U, where you get the speed, comes right down. We have seen that U go down to quite a narrow V. We did some work some years ago on the connector, the I-75/I-85 connector through the center of Atlanta. The results were so impressive that we decided to apply it metro-wide and now have something like 145 ramp meters in place and will be up to something over 160. We think, in most cases, they do provide a reduction in the hours of congestion, but it doesn't eliminate congestion.
Q. Your performance metrics are freeway-centric and driven toward freeway incidents, level of congestion, flow dynamics. How do you intend to extend your performance metrics down to the municipalities? That's the first question. Does your regional concept of operations intend to move in that direction as your system or connectivity matures to encompass and interface to traffic signal operations? Here in Hampton Roads we have numerous municipalities that have state of the art traffic management systems geared toward traffic signal operations and arterial traffic. How would you envision merging these?
A. Marcus Wittich: The first part of this, of course, this specific project has not considered at all, this narrow project has not considered arterials.
Hugh Colton: Or city and county.
Marcus Wittich: Right, or city. I could certainly envision somewhere down the road that we will look and address that as is appropriate. But at this time, we haven't given it any thought. Now, the question also opens up to a larger one with relation to GDOT cities and counties. Hugh, I don't know if you want to talk to that.
Hugh Colton: Well, we plan to replace our ITS system called Navigator in the very near future. We are going to make sure that there are parts of that which really can be used by the transportation control centers that the City of Atlanta and that the counties have. There are several of them already in use in the Atlanta area who take care of the City of Atlanta streets and the county streets. They are very proactive in using the common CCTV system that we have. They can add their cameras to our joint system, so they are able to monitor the traffic on the arterials and on the city and county streets and make traffic signal timing changes on the fly. We plan to put that into a future system. Obviously, when it's in the system, we will have to have some way of reflecting that performance in our performance measures. That is not actually part of what we're doing at the moment. That is something for the future, which we certainly intend to look into.
Q. Are you measuring travel time by time of the day, day of the week, and routes taken? If so, would this data be made available to the public?
A. Hugh Colton: This data is available on our website today.
Q. Will the first iteration dashboard be available on the web?
A. Marcus Wittich: No. It's internal only.
Q. How are you working with or planning to work with the metropolitan planning organizations, MPOs, or the local agencies to develop multidirectional performance measures?
A. Marcus Wittich: This is not currently part of the scope of this project.
Q. Do you think that tracking performance measures of ITS will help for more sustainability and reduce citizen's carbon footprint?
A. Marcus Wittich: Absolutely. The intent of these performance measures is to help us identify directly how our activities impact congestion and accidents and reduction in congestions and accidents reduce carbon footprints. The whole purpose of what we're doing is to help us reduce the carbon footprint. Ultimately, that's one of the goals.
Q. In the presentation, it lists one of your performance metrics as incidents. However, you state verbally that it is accidents. We see a vast difference in that meaning, in the terminology. Is this performance metric solely accident or does it involve other incident types as well?
A. Marcus Wittich: The answer is we weren't talking about a specifically narrowly defined performance metric, but a category. The questioner is absolutely right—there is a big difference between incidents and accidents. I think the metric that we're interested—probably is there will be several metrics that will cover that. I think we will certainly cover accidents, and probably different levels of severity, because those tend to have a significant impact on congestion. We will also be looking at the other components that make up incidents, because those also have an impact. But we haven't defined a specific performance measure or suite of outcome performance measures for that, it's just a category.
Q. Are your VMS portal or permanent ones?
A. Lokesh Hebbani: I think the ones in this presentation are permanent.
Hugh Colton: Yes, yes. We have permanent ones. We do have portable ones, but our portable ones are usually used for planned construction projects and are taken out to site-most of them are, in fact, operated by the contractor who is doing the construction or the repair project.
Q. The question is: What is your approximate annual budget for developing and maintaining the performance measures? Don't include people involved in ITS operations, like full time employees.
A. Marcus Wittich: That's an interesting one, because we actually don't have any specific budget for this. The answer to that would be zero. However, what we are doing is using pieces and parts of our ITS operations budget that is currently used to produce our weekly performance measures report and we're redirecting that. We also have had some support with GDOT IT and they are providing us manpower, but we don't have a specific budget on that either. We're on a shoestring.
Hugh Colton: But by mechanizing a lot of the data collection, we are in fact reducing the number of man hours that are necessary to collect the data. So that is really giving us a performance improvement and therefore, we have time to do some of these other activities using the same labor.
Q. What type of database does this web-based dashboard use? Is it Access or Excel or Sequel?
A. Marcus Wittich: The GDOT enterprise architecture is a Sharepoint foundation. They're building the user interface on top of Sharepoint. The database underneath it I believe is going to be Oracle. Again, I don't have much visibility into that and actually don't care, because that's the IT folks. We're also leveraging the GIS system, the enterprise-wide system which is ESRI. Those are the three major components.
Q. Comment from remote audience: The Missouri DOT has a 24/7 customer call center and is effectively using a survey process to quantify customer perception and is available on their website. In there are performance measures also.
A. Marcus Wittich: Fabulous. Thank you very much, Missouri DOT. We will love to look at that and see what we can leverage. Thank you.
Q. Have you looked into the technology that tracks Bluetooth MAC addresses as a way of measuring travel time?
A. Hugh Colton: No, but we have used cell phone probes over an extensive area. We have had some good results and we have had some bad results. We haven't looked into some of these other technologies like Bluetooth, etc., because this means we have to get detection equipment up close to the freeway. By using cell phone probes, we can just go to a reseller of that data and they'll take care of collecting that data, which is probably a much more cost effective manner of doing it. We don't have to apply any hardware in the field. We don't have any maintenance costs doing it that way.
Q. Is there more written information available on this project other than this PowerPoint presentation?
A. Marcus Wittich: We have a concept of operations document. It hasn't been published yet to GDOT. I expect that to be released, although it's a work in progress so it will continue to be updated. But I expect a version of that to be released in the next month. We certainly can make that available through FHWA to anybody who's interested in it.
Hugh Colton: Once that is released, it goes on a database under the Freedom of Information Act. We just supply it. In fact, the configuration manager, who's sitting in the room here, is nodding his head that it will be available. Give us a couple of months to get it in there and you can have a copy.
Q. Marcus, what's your background? Are you in management or a statistician or an engineer?
A. Marcus Wittich: Let's see. Undergraduate, I studied math and statistics, as well as economics and computer science. I spent many years in management consulting as well as information technology consulting to the private sector. I've been working in the public sector in the ITS arena for now nine years. I've also run my own business. I spent ten years building a printing and prepress business. So I think it's a little bit of everything.
Q. You talked about contemplating the use of regression analysis in the future. What casual relationships do you wish to explore?
A. Marcus Wittich: When you think in terms of congestion, it's looking and isolating. In a regression analysis, you want to look at what are the inputs. There are very many things that impact congestion. Ideally, what we want to do is isolate so that we can see the impact of what we do at the traffic management center. A lot of people have told me that I'm dreaming, that that will never be possible. Ultimately, we want to isolate out all the major impacts of delays. So what could that be? Weather is certainly very obvious. Accidents impact congestion. There's actually a relationship between both of those.
Hugh Colton: Another important thing is getting the lanes clear. We have incident management teams set up with the local jurisdictions and it's a slow process. But our colleague who heads up the HEROs, the Highway Emergency Response Operatives, has worked a lot with these people so they can understand what the impacts are when they arrive at the scene of an incident. Do you need to park the fire vehicle across four lanes when, in fact, blocking two lanes would be sufficient? I know they want to provide more protection to their workforce, to their firefighters, but there's got to be a limit to this. By showing them the right way to do these things, to give them the best protection, we can actually get more lanes open. We're very fortunate here in the Atlanta metropolitan area we have no freeways with less than four lanes. In some places we'll have as many as nine lanes. So if we can reduce the effect to just blockage of two or three or four lanes, we can still keep traffic going. But if every time there's an incident the state patrol turns up and wants to close every lane, then yes, you cause greater problems. So we work with them to reduce the impact. As I explained earlier with our TRIP program, we have greatly reduced the impact of major truck incidents. That's what we're working on.
Q. Are results from the testing of Bluetooth MAC-based travel times measurements available? But as Hugh mentioned, I don't think we are using Bluetooth.
A. Marcus Wittich: No. We're not using Bluetooth.
Comment from remote audience: Washington State DOT is working with the University of Washington to research the feasibility of the Bluetooth technology for travel times. Some initial analysis is available. Send an email to BaileyPE@wsdot.wa.gov for more information. Hugh, you can get more information from them.
Q. To what extent do you have higher education support through universities or the research community?
A. Marcus Wittich: Actually, right now we have one guy in IT and we are actually working with a professor at Georgia Tech, and using a lot of his algorithms. Just to readdress the regression analysis comment from before, I think one of our targets is to engage a PhD student in this type of work. It's the perfect type of work for academia to look into and explore. We have not yet made that contract, but we certainly have started exploring it and talking to folks to look for something like that.
Hugh Colton: If I could add one thing about clearing incidents quickly, let me just copy some information to say that with our TRIP incidents we are saving $300,000 per TRIP incident by the reduction in the delay, that's calls to everybody, fuel saved, and emissions problems.
Ron Beard: This is Ron Beard with the Georgia DOT. That's a conservative number and it doesn't even include all the different things that are saved. That's the most conservative number that we have at the moment.
Q. As the Department of Transportation is moving towards performance-based contracting, do you see any challenges in terms of data collection, and focus of study?
A. Marcus Wittich: I'm not quite sure what they mean by the second half, but when you bring in performance-based contracting, although I didn't include it in this presentation, one of Serco's personal motivations in doing this process is to get significant and tested performance-based measurements so that the DOT, if they so choose, when our contract is being renewed to let that contract as a performance-based contract. We are very much interested and have done this a lot in the UK. We are very much interested in the United States in performance-based contracting. We want to get paid when we perform. We want to make more money when we perform and we don't want to get paid if we don't perform. We believe strongly in that. So this effort is a foundation to allow the Department, if they so choose, to be able to do that and measure it effectively.
Hugh Colton: If you want to contact the UK equivalent of the DOT for England, they have a contract with Serco to run their national transportation center just outside Birmingham and they are on a performance-based contract. Initially, it cost Serco a great deal of money, because the performance bar was set so high initially. They really had to excel themselves to get up there and stop having major penalty points every month. It was costing them $500 pounds per point and one month they had 200 points against them. So, 200 times $500 pounds was the amount that was deducted from their monthly invoice at one point. It gives you some idea of how you can really force performance-based contracts on contractors and get them to perform.
Comment from remote audience: The University of Portland has a contract with the Oregon DOT and City of Portland for an intern involvement that's like a grad student as an extension of staff. Professor Robert Balkani can provide the information. Denver, Colorado has a like contract with the University of Colorado. These are available to view or use as examples.
Marcus Wittich: Thank you very much. That's good information.
Q. Have you run into any of the privacy issues on cell phone data?
A. Hugh Colton: I think we have demonstrated well enough the anonymity of this information. All we receive is the speed of vehicles. That's the only information we get. We don't know what the caller ID was or anything. We just get an X-amount feed that comes from a contractor.
Marcus Wittich: It's an aggregate speed. It's an average. We don't get any individual vehicle cell data. I think the way the companies we've been working with get around that is the computers inside of the cell phone company strips out any private information and only feeds them anonymous data. The cell phone data generally is not tracked by GPS. It's actually tracking cell phone handoffs between towers and using a statistical methodology to correlate what the probability is of what roadway that that group of traffic is traveling on. It's really quite abstracted from the privacy issue. I don't think that technology really has that issue or fear.
