Using License Plate Readers to Fight Crime, Part 3 of 4, Dale Stockton, Program Manager, Road Runner, Automated Regional Justice Information System

Dale Stockton: Good afternoon. I'm Dale Stockton. I work with the Automated Regional Justice Information System in San Diego, and by way of introduction, I spent more than 30 years carrying a badge and a gun, so I'm definitely the practitioner side of this. I'm not a technologist, but I have to tell you as an up front this is the best tool we've had come along in a very long time. It can really impact crime. It can really impact public safety, and I want to share from a practitioner perspective how this is all beginning to come together in the field.

Briefly, as introduction, license plate readers today — and it wasn't always this way — but they're much more effective and they can actually capture a plate at a closure speed of 150 miles an hour; in other words, two cars coming at each other, that plate reader is able to pick out that plate and check it. And although it's often obtained for using for fighting stolen cars, the reality is we're beginning to understand that the value goes far beyond stolen cars, particularly in the area of investigative support, but it can also be used to check any database; for instance, your local registered sex offenders.

Now, it's important to realize, though, that the checks are being made against a static database, not a live database, because the reality of LPR is it can run thousands of checks in a day, one unit can, and if we were to be doing that against a live database, such as shipping it through inlets, it would literally bring the system down. So you're only as good as your last download of information.

Now, once that plate is read, that plate is retained, and assuming the agency stores the data, it can later be subject to a query for investigative purposes. The biggest reason that we want to use LPR in the field, though, is that it truly is a force multiplier. Officers can do their normal job on patrol, and that license plate reader unit can function in the background while the officer can pay attention to what's going on around them.

Now, what we know is that a good officer, a very proficient officer, can usually do somewhere in the neighborhood of 100 to 150 checks in a day if they actually can dedicate themselves to that in addition to the regular patrol duties. A license plate reader can do 5,000 checks, a lot of variations depending on how much traffic is going by and the speed of the vehicles, but suffice to say thousands can be done in a day with a license plate reader.

And, as Meghann mentioned, the value of the data is that it provides you with a photo, a time and date stamp, and a GPS location of where that plate was seen. So, when you start thinking about how that plate can be used, the investigative value is really at the top of the list. It also helps you when you're trying to find somebody, not only suspects, but perhaps a witness that has drifted off the radar for whatever reason and is essential to a trial, you can look and see where that person's vehicle was last seen, and that gives you a place to start.

One of the more recently realized applications is in support of homeland security because, around this country, we have identified infrastructure that we know is at risk, and one thing about that infrastructure: It's not moving around. So, if we start looking at vehicles frequenting that infrastructure, that can give us greater insight as to where potential risks are.

And the way I like to put it is, essentially, you can tell which vehicle is where, when. Notice that's a retrospective look, which vehicle was where, when. Give you a real-life example: Escondido, California; this is an agency in northern San Diego County. They had a hit-and-run case. It was a fatal with a pedestrian. They didn't have a lot of information. All the witness could say is a partial plate and a gray Ford Taurus.

Now, in California, those three missing letters would equate to thousands and thousands of possibilities, and quite frankly, with DMV, it just wasn't of any use. They checked some normal resources and they couldn't come up with anything, and the interesting thing about Escondido was they had one LPR unit, a single LPR unit on a car, and they had only been using it for a very few months.

The unit was originally obtained for the purpose of finding stolen cars. So a couple of officers, shortly after they got the unit, got the permission of a supervisor, and they came in. They were going to catch the big one. They wanted an occupied stolen. They got the LPR car. They go up and down every street in town all day, and at the end of the day, they had gotten absolutely nothing. And they are pretty, pretty disappointed about it. To be candid with you, they felt like, “Gee, you know, this thing didn't work.”

Well, unbeknownst to them, what they had done was they had captured an LPR read on a gray Ford Taurus with that plate, and that's the actual capture there. The picture is not real clear, but trust me, that's a gray Ford Taurus, and they had the full plate number on it. And, of course, with an LPR capture, they were able to map it, and that mapping led to the officers going out, checking the area, subsequently finding the vehicle and the guy actually working on the car, trying to repair the damage. So those officers that had diligently worked their town looking for a stolen car had unknowingly solved a crime that wasn't going to occur for another three months.

Now, like most data, the value is when you can look at it on a larger scale, and to do that with LPR, we've really got to share the data. That's really the key to making it happen. This is just beginning to happen around the country, but it needs to happen a lot more. I'm a big, big proponent of this, and without it, we simply have information silos sitting around different agencies, and the value to individual agencies within just their jurisdiction is really limited.

So the key here is linking agency reps early and linking them often and having them share the information, and ideally, you'd have those agencies sharing data because not only does it give them a more effective alert and analysis of the data, but it lets them have some degree of economy of scale when they go to make purchases. It also gives them better vendor support, and what we found is that lessons in one agency are usually applicable to another. The learning curve on LPR is not real steep, but certainly by learning from others, you can get there a lot more quickly.

And then, finally, if it's suitable for the region, you can even share infrastructure, such as a server where data can be pooled and queried by multiple agencies, and I will give you an example of that here in just a minute.

In the San Diego region, which is the area I am primarily familiar with, we have both mobile — well, I won't say both. We have mobile LPR, we have portable LPR, and we have fixed LPR, and those are the agencies that are currently using those different types of license plate readers; however, in spite of the Regional Justice Information System, RJIS, and our long history of sharing data among the agencies, when it came to license plate readers, it was unique to each agency. We were creating our own information silos.

So, as a result, they were each holding their own data, and virtually no sharing was going on. When we looked at that, we just said, “This is not the way we do business in this region,” and we wanted to do it differently, and the solution we came up with was a regional server.

So, in making that decision, we considered, you know, what would be the advantages, and certainly one of them was by providing that data storage. License plate readers generate a huge amount of data, and if you don't have significant IT infrastructure, it can overwhelm you pretty quickly.

It also, obviously, facilitates the sharing of the data. You're pooling a much broader region, and you have a better grasp of what's happening in your area, where those vehicles can be located.

It permits query by non-LPR-using agencies, and this is one that's often overlooked, but it's really important. I want to emphasize this. There's a lot of agencies. You see those patches down at the bottom. Well, those are the agencies in San Diego County that have marked patrol units and do typical municipal policing-type functions, but RJIS has 82 member agencies. So, within the San Diego region, there are 82 justice-related agencies that would really like to have access to that data, and because they don't have license plate reader systems, absent a regional server, they don't have it. By building a regional server, we give them access to that data.

It also provides the storage option for new LPR agencies. So an agency that's looking to get into license plate readers can dedicate more of the money towards the hardware that obtains the image as opposed to building the IT infrastructure that's necessary to store the data.

This type of approach also supports a regional alerting function and encourages agencies to develop a regional hit list, hot list, alert list, whatever you want to call it.

And finally, longer term, it gives you an infrastructure on which to build. We are building out our system. We have just started with fixed. We have two license plate readers that are fixed in the county currently. We'll be adding another two dozen, maybe three dozen by the end of next year. Now, fixed readers generate a huge amount of data, and absent that regional server, I'm not quite sure what we would be doing with the data.

So the status of our project: As of now, we've got the regional server. We got the regional server as a result of a stalled federal project, but, unfortunately, when we started we had no software to handle the data. And then came along NIJ with a project they'd been working on. They funded SPAWAR Charleston to develop a standards-facilitating software. It's called LPRD for LPR data, and Mr. Ford, bless his heart, agreed to let us use that software to support a regional server, and it's working. The intent was to provide for support for both a multi agency and a multi vendor effort.

As far as I know, this hadn't been done anywhere in the country, to combine both supporting multi-agencies and to be vendor agnostic, and as of today, it's operational with three agencies. We are adding more. We wanted to mature the software a little bit and make sure that it was as stable as could be, and it's also working with multiple vendors. Currently, the software works with ELSAG, NDI, PIPS, PlateScan, and we hope soon it will be working with AutoVu. Now, if your favorite vendor isn't listed there, I'd encourage you to encourage them to find out about the standards effort, because one of the things we heard from the vendors at the front end was, “Well, if the customers ask for it, then we'll do it,” so they need to hear it from you. And along the way we've tried to encourage them by noting that law enforcement will be looking at these standards when they go forward in terms of grant purchases.

So, in terms of giving you a visual on it, in San Diego, in essence, what we have is multiple agencies feeding into a centralized regional server, and by so doing, each agency can retain its own data, setting its own retention standards, and they're not limited to one brand of license plate reader to buy.

Additionally, because of the access to the data by non-LPR-using agencies, any RJIS member can query the data, and if you're wondering what kind of agency that might be, it could be probation, it could be a district attorney investigator's office, social services. There are a number of agencies that benefit from criminal justice data.

Now, let's take a look at a different approach. I'm not going to say this one is better or worse. It's different, and it's significant that we look at different systems because the reality is we've got different systems springing up all over the country. This is the Los Angeles approach. They've got more than 25 agencies; they're sharing, just beginning to share data now.

Now, their data is able to be shared because they're using one vendor, and that vendor facilitates a Web query-type sharing, and each agency in that case is holding its own data. You're relying on those agencies to retain the data for you to be able to access it, but for them, it's working. And it depends on where you are in the rollout cycle, which approach might be better for you, but my thought was, well, we want a system of systems; we want to be able to share with San Diego, between San Diego and Los Angeles, and we have been actively discussing this with them to make that happen. We want our agencies to be able to query L.A. data and, conversely, L.A. agencies to be able to query San Diego data.

Now, what's important about this is that we have a system of standards, because if the data isn't in some type of mean compliant Global Justice XML standard, we are not going to be able to make that happen. And vendors, bless their hearts, they bring out their products, and they tend to be somewhat proprietary. They want you to buy more of what they have to sell, understandably, but so far, the cooperation level on the part of the vendors has been pretty good, and they are moving forward with the idea that standardized data will be something they'll support.

So, as we go forward on this, I showed you the L.A. and the San Diego thought, but there is a great deal of discussion around how do we do this on a national basis, what are we going to be looking at as we move forward, and specifically, we believe — I believe — and there are some people that are working on this with me — that the key here is we develop regions around the country.

We're probably not going to have any centralized national giant bucket of license plate reader data. It probably wouldn't stand the court of public opinion, and it's probably something that given where we are in the rollout cycle wouldn't easily be done, but we can develop regional sharing capability, such as the San Diego effort and the L.A. effort I showed you.

And so doing, you get those set up and then begin to share between those regions, and as you begin to look beyond your region, utilize a trusted broker like Nlets, and if you're not familiar with Nlets, they have a booth set up at the front. I'd encourage you to check with them, but they're a trusted broker. Every law enforcement agency has a connection to Nlets. Nlets would serve not as a storage unit but as a pointer system, something akin to a Google, so that when you check a plate, Nlets would point you in the direction of where that plate can be found, and the result of that would be a query in one state by an investigator could give an indication of plates of interest in other states, and then that information can be pulled back.

Now, we're not there yet. We're not anywhere close to that yet, and yet we're standing up these systems. So I strongly encourage anyone involved in LPR in the interim to establish an e mail group and do an e mail blast when you have a vehicle of interest. This is working in the southwest area of the United States. I've seen these go out on average about once every two weeks where an agency has a high-profile crime, sends out their e mail list, and several times, within just an hour or so, will have a significant lead on a vehicle that they're looking for. It does work.

One of the questions about license plate readers is this kind of hocus pocus, “you are invading my privacy; it's super intrusive.” And I come from the opposite end of the spectrum. I truly believe that this technology is only doing what license plate readers have already — I'm sorry — what officers have already been doing in the field for many, many years.

The courts have indicated to us that officers can look at a plate and run it. So let's just go down quickly here through the two sides, an officer doing it and the license plate reader doing it. Officers can run plates anytime they want. The courts have held that's why we put the state plate on there, so an officer can check and make sure it's current and it's not wanted. Well, the license plate reader is doing the exact same thing. It's looking at it; it's checking that plate to see if it's wanted. The officer can pick and choose among the vehicles that he or she looks at.

Now, this is an interesting differentiation because license plate readers, on the other hand, they don't differentiate. Some people would say they're not biased, and to any of you concerned about any type of bias-based policing, license plate readers are absolutely the antithesis to racial profiling. They don't care what color you are, what color your car is. They're picking up all those plates.

The officer can write down and record that info, and officers frequently do when they see something a little out of the norm. They'll jot it down, jot the time down, and that's what a license plate reader is doing all the time, noting time, date and location.

The officer can hold onto that info, and they do with their memo pad until such time as a detective comes down to briefing and says, “Hey, does anybody have anything on this type of information?” And, of course, the LPR reader is also holding that data.

An officer can provide that information to an investigator upon request, and a license plate reader can be queried for that same information. Past information may be very relevant to a crime later discovered; in other words, “Yeah, I saw this three days ago, and maybe it's something you're looking for.” And obviously, the license plate reader can provide that same retrospective look.

Now, the difference is on the officer's side, there's a weakness. The officer has to remember it, write it down, has to be aware of the request and then able to provide it to the investigator. So let's change gears here for a minute. I want to provide you with some of the myths in LPR and dispel them.

Specifically, “my vendor is best.” There's several good vendors out there, probably half a dozen major players, and it really depends on your needs, and I always ask people what's the area you're in, already using. That always helps.

“LPRs are primarily for finding stolen cars.” We've already seen examples of where that's not the case. We're certainly getting beyond that.

“They take the human element out of policing.” One of the things we often counsel people about is you can't depend on the information that just pops up on your screen. It's not a death ray. You have to confirm the information and take appropriate action like any responsible officer.

“They target people unfairly.” Well, as I just mentioned to you, they're absolutely color-blind when it comes to plates and cars and people. They pick up everything.

“They are too expensive for small agencies.” The good news is the price is coming down. The officer activity … you'll hear about that in the next presentation, but suffice to say, it's truly a force multiplier.

Here is one: “The accuracy rate is this big.” And this is primarily a vendor-driven thing. I'll give you this quick example. They'll give you a number, 95 percent accuracy. Well, here's what you need to ask to be an informed consumer. It's what's the capture rate, what's the accuracy rate. And I know we have a lot of academicians here. Here's a quick example: 100 cars in a parking lot; 50 of them get read; 49 of them get read correctly. To a vendor, that's a 98 percent accuracy rate. That's not 98 percent to me, but to a vendor, that's 98 percent because of those 49 it got right, 49 out of 50, that's 98 percent. Well, I'm thinking 49 out of 100.

“My cameras read all 50 states.” The reality is cameras have to be configured for the region of the country that you're in, and while they may be able to be configured for all 50 states, don't let them tell you they'll read all 50 states as if 50 different states could go by and it will differentiate between them.

The hot lists are only as good as you make them. The alerts are in need of being updated frequently, and as Meghann said earlier, wireless uploads, real-time uploads are important.

This one is going around a lot now. “If you have got an old computer, just add a camera and you're good to go.” It's a lot more than that. The algorithms are important, and watch at the next IACP show. You'll see some brand-new vendors.

Bottom line: Do your homework. Make sure the vendor has some experience in your area.

I've got some tips for you, again, from these trips we made around the country. I have to go through them quickly in the interests of time, but, first of all, wild cards, very, very important, and I got to give you this story to show you the importance of it.

More than 30 rapes in three years in an area of Florida; the suspect would choke his victims into unconsciousness before he left. As a result, there was no vehicle information until they finally had a break where a victim was able to give a partial plate. The analyst in this with just that partial plate, even though one of the numbers was wrong, was able to get a suspect vehicle in a matter of minutes.

They used that information. They set up surveillance, and they actually targeted the guy, apprehended the guy as he was in the process of going after another victim. So remember Victimology 101. Meghann told you the story about Jefferson Parish; that's a great example, you know, where was your victim. It's not just where the suspect is; it's where the victims are.

Hot list, clear instructions as to what to do. Hot lists need to be updated. The more frequent you can do it, the better. Real time is key. Share, share, share to be effective. The broader perspective you have, the better. Hot spots, hit them relentlessly before, during and after an event, because if you have something go wrong, you'll have a resource of information to go back to in terms of people to look at. You might just bust an alibi.

Utilize wireless upload or download, if possible. Many agencies are still using USB, but cellular makes a great backhaul to do this.

Accuracy in time, accuracy of both time and GPS data. This is a new evolving issue I don't have time to get into sufficiently now, but suffice to say, the time and the date and the location stamp have to be accurate on that LPR read if we're going to utilize it in any type of investigative endeavor.

Keep the LPR in the field. If it's not in the field, it's not capturing LPR information; it's useless. So think about that when you are assigning the units.

And finally, registration enforcement. It's time for a new perspective. We've got to put an emphasis on getting those plates displayed cleanly, so the LPRs can read them.

Maybe I have one more on here. Yep. This is a freebie. If you have LPR units, don't run them through the automatic car washes because over time, guess what, the camera doesn't point in the right direction.

That concludes my presentation. That is my contact information.

[Applause.]