With our safer roads mandates, why does traffic data volume, coverage, and accuracy matter?

In this video from Arity’s session at ITS America 2023, you’ll learn:

• What new data sources have emerged?
• How can this driving data help the public sector achieve Vision Zero goals?
• How does an agency implement these sources?

Speakers:

• Dr. Sisinnio Concas, Program Director, Center for Urban Transportation Research
• John Hansen, Senior Engineer, Olsson Engineering
• Anthony Johnson, Senior Sales Engineer, Arity
• Erik Dietz, COO, Michelin Mobility Intelligence (MMI)
• Sandi Dunmyer, Business Development Manager, Arity (moderator)

Transcript

Sandi Dunmyer, Arity:
Well, hello everybody. Thanks for joining us. Does that sound okay? Okay. I’m going to start by answering a couple questions. You might be wondering who is Arity? Why are we doing this? So Arity is a mobility data and analytics company. We’re a wholly owned subsidiary of Allstate Insurance, but we are our own separate entity. Arity has the largest driving behavior database informed by decades of understanding insurance risk collected in large part through telematics first party data via popular phone apps. So Mobility Intelligence Insights is being used to produce the most complete picture of road safety with its all-inclusive look at all roads without bias. So we can talk a lot more about that during our discussion. It’s collecting data from any driver on the road that has a cell phone. So is there anybody here that doesn’t have a cell phone? No. Okay. So you get my point, right?

The current count of U.S. active connections is 30 million plus. So that’s a very, very large scale data. Today, Arity is hosting this session because we want to discuss data as it pertains to safer roads. I think we all agree it’s imperative. We’re all involved in these discussions, and we’ll have time for Q&A and further talk from our experts after we share some thoughts. Vision Zero‘s an extremely bold, worthy goal that began in 2014 to end traffic fatalities on our roads. Unfortunately, we’re at a time that U.S. fatalities have hit an unacceptable 20 year high.

Today we’ll concentrate on what’s available to aid our assessments and our solutions as we seek to accurately identify and equitably prioritize the impact of infrastructure on crashes. What we’re here for is, how does the data type, the volume, and the frequency of data matter when looking to identify the specific causes of these fatalities and these accidents? So over the next 45 minutes, I’ll host an informal discussion between four experts to shed light on the challenges and opportunities for a variety of use cases. And again, we’ll have Q&A at the end, so make sure you have some good questions. Ready for these guys? The top three things we hope that you’ll leave with are one, when it comes to new data sources, where have we been and where are we going? Two, what does this data mean to us in regards to safer road mandates and the Vision Zero goals? And three, how does an agency implement these sources?

And now for our panelists of experts, as I introduce each of you, I’ll also ask you a question to help the audience get to know you. Dr. Sisinnio Concas serves as program director at the Center for Urban Transportation Research. He’s the founder and director of the Autonomous and Connected Mobility Evaluation program, leveraging his 20 plus years of experience as a transportation economist. His research focuses on the evaluation of novel and disrupting technologies. I love that title, novel and disrupting technologies. Dr. Concas holds a PhD. He has a doctoral degree with a field specialization in macroeconomics from a very prestigious Italian university. I’m not going to try to say the name. He currently serves on the TRB standing committee and a subcommittee. And a fun fact about Dr. Concas is that transportation runs in the family. His father, his grandfather, and his great-grandfather were all born in rail stations in Sardinia, Italy. So first question I’m going to ask each of these guys as I introduce them is as a stakeholder, Dr. Concas in the Safer Roads mandates, what is your 30-second viewpoint on data as it regards to safety and mobility on all roads?

Dr. Sisinnio Concas:
Well, thank you for the invitation. I think this will be a great discussion in 30 seconds. I would say timely data, accurate data, representative data, and I think we will discuss what representative means that the timely availability of data, the frequency at which we might need data. That’s it.

Sandi Dunmyer, Arity:
Excellent. I thought with academia, I wasn’t sure if I was going to get 30 seconds. Good job. We have John Hansen at the far end there from Olsson Engineering, subject matter expert in transportation. John has 41 years in the transportation industry. He’s steadily supported ITS for 22 years. Two of those as president and has been a member of the National Committee on Uniform Traffic Control devices for 22 years with eight of those as a delegate. He’s also heavily involved with Rocky Mountain ITS as the vice president and several IT committees and relevant task forces. So fun fact for John, my whole career has been associated with the enhancement of our highway system and safety. I’ve logged over 2.75 million commercial airline miles flying towards that achievement. That’s pretty great. So same question to you John, as a stakeholder in the transportation family, what is your 30-second viewpoint as it regards to safety and mobility on all roads?

John Hansen:
Well, I’d like to say thank you to Sandy and the team that’s here and welcome you. I really represent kind of the infrastructure, as you can tell by my bio to talk about where we are and where we’re going. You almost have to talk about the past. And the past doesn’t reflect that we’ve had the data elements that we really needed to adequately address the safety mobility issues, but all of a sudden we’re faced with unbelievable amounts of quality data real time. And now it’s a matter of the management, the structure, the architecture, the standards, and building on that to bring that back to the entrepreneurs and the practitioners will be using that information.

Sandi Dunmyer, Arity:
Great, thank you. Next is Anthony Johnson. Anthony is an Arity senior sales engineer and has over 20 years of experience in geospatial analytics. He holds a BA in geography from the University of Texas at Austin and an MBA from Texas Christian University. He has a passion for all things spatial, especially when it involves the opportunity to improve road safety and save lives. Anthony, fun fact, I had to make one up because he didn’t send it. So watch out. He designs amazing custom sneakers for all types of clients in his spare time. And these can take what, 30 to 50 hours and more, right? So pretty interesting. You’ll have to ask him for some pictures. What are you wearing right now? Where’s Sandy’s customers? Yeah, he doesn’t have any on today. Anthony, same question. Would you like me to repeat it? As a stakeholder in the transportation family, what is your 30-second viewpoint as it regards to safety and mobility on all roads?

Anthony Johnson, Arity:
Thank you, Sandy. I appreciate the opportunity to be here and what I would say is a few different things. First of all, it’s an opportunity that we can make a contribution to Vision Zero being in this data space. What I would also say is because it is new, what’s top of mind for me is education, right? Because we’re bringing in this new technology, this new data, that whole fear of the unknown. So as we bring this technology forward, what’s top of mind for me is how do we best educate and inform this space on what it is and what can the data do?

Sandi Dunmyer, Arity:
Great. Absolutely agree with that 100%. Thank you. Erik Dietz is next from Michelin Mobility Intelligence. Erik serves as the COO for Michelin, MMI, I’m going to say as they deliver technology driven mobility service in multiple regions that serve the public sector and B2B customers. He has a bachelor degree, a master’s degree, and is currently pursuing executive education programs at Harvard Business School. He’s passionate about technology that makes people’s lives safer and more equitable while lessening the impact on the environment. So fun fact for Erik as he is driven across all lower 48 states, including 10 road trips, coast to coast and back again, it’s a lot of miles. So Erik, as a stakeholder in the transportation family, what is your 30-second viewpoint as it regards to safety and mobility on all roads?

Erik Dietz:
Thank you, Sandy. I think as everyone here has said, happy to be here talking about what we all know is a very important topic for me. I think transportation, especially for Michelin, is really a moral imperative for us to improve the safety, the equity, the sustainability of mobility on our roads. And I think today we’re here talking about data and I think there’s a lot of data that’s out there that we can talk about, but rather than focusing only on the data, it’s really, as we’ve said, the quality of the data and what types of data, how are you fusing that with other data types to really produce those actionable insights. So when we talk about education, I think that is absolutely the first step, but then how do you implement what we are learning, what we are extracting from this data to make that measurable difference on our roads?

Sandi Dunmyer, Arity:
Excellent. Awesome. Okay. Probably a lot of us are familiar with the National Highway Traffic Safety Admin and the stats that we’re seeing that are not favorable there were just under 32,000 fatalities on the road in the first nine months of 2022. This has estimated to rise to 46,000 for 2022. That’s staggering. I know nobody in this room is okay with these numbers. I think most of us will also agree. It’s obvious we’re missing something very important in our safety strategies. You don’t know what you don’t know. So we’re going to try to uncover some of this. First question to Dr. Concas and John first, how do we uncover what we don’t know using data as one option? Dr. Concas, you want to go first?

Dr. Sisinnio Concas:
Yeah, definitely a difficult question because at least speaking as an academic researcher, we always look at what’s out there currently available in terms of the data. And I think about in terms of safety based on those dire statistics that you just presented to us. What we’re phase two when we are trying to understand some very important questions about causality? For example, I live and work in the state of Florida, which is a very fast growing state where definitely safety is paramount not just to the Department of Transportation at the state level, but it really affects the livelihood of all of us. So as a researcher, when we study, for example, the crash analytics, the data sources that we work on, at least in the state of Florida, is already crushed reported data. There is a timely process of reporting data. For example, in the state of Florida, we have access to geospatial information about each one of the crash events, and that allows us to conduct some thorough analysis, but really to get into the causality aspects.

Really what are causing certain type of crashes is that due to infrastructure, is that due to travel behavior? And if so is travel behavior and therefore the crash is affected by the infrastructure by ourselves. And that really requires having access to more fine grained data about how people behave. Now we get into the realm of conflict assessment and we know that there is a challenge in terms of understanding how can we get data that can help us to understand precursor to crashes that we already observe in these existing databases. And there is also a problem of representativeness in the databases that I hope we can discuss. And actually representativeness also deals with issues related to equity in terms of the information at our disposal. So there are needs and challenges I hope we can share and shed some light upon.

John Hansen:
Thank you. I’m going to provide maybe just a window into a real world situation that we currently deal with and then we can talk about the impact that this is going to have around that. The issue is, well, zero fatalities, we’re all familiar and we understand what that is about half of the fatalities we have on our highways occur in rural areas. The average response time to an accident in a rural area is greater than an hour. That is the magic hour that you always hear about that can make the difference in a life. So with the advent of this kind of information, the quality of it, we envision the opportunity given the right circumstances, information and criteria, we can almost predict where we need to be aware of these circumstances and look at the data to help us in the prediction side of an accident based around that criteria. So that’s a real world example. And if I had EMS and first responders and others here, they will tell you that is so critical to what their mission is in that area.

Sandi Dunmyer, Arity:
Great. Thank you Anthony. Erik, anything you want to add to that one?

John Hansen:
No, I think that covers it.

Sandi Dunmyer, Arity:
Okay, good. Okay. Tell us which types of data you rely on most and why and which types of data you feel are you’re missing that you think would be valuable? John, do you want to add that one?

John Hansen:
We’re going to have you repeat.

Sandi Dunmyer, Arity:
Sorry. Oh, sorry. Tell us which types of data you rely on most and why and which types of data you feel like you’re missing but you think would be valuable.

John Hansen:
Most of the data we have right now is post-process information. A lot of it’s collected well in the past and it’s just the nature of the data we have to do the kind of information sharing we need to have. This represents a total change to that. So first of all, it’s going to be fresher, it’s going to be new. It may even be real time, but the quality of the data, the amount of data is what’s really exciting here. You’re going to hear today about where we can collect that. I had a discussion with one individual while here that in Europe they have about 30% of their cars that they can actually capture live information from. In the US we’re around 17%, but just think about that alone. If we had live information from vehicles real time, what an impact that’s going to have in all that we do around transportation. So hopefully the takeaway from this today and from this conference is the huge impact the real time fresh data is now going to have for transportation and how it affects ITS,

Sandi Dunmyer, Arity:
Dr. Concas, anything to add?

Dr. Sisinnio Concas:
Yeah, I’m going to, the question is what kind of data do you rely the most? And I am trying to split it in two basic areas of needs. Currently I have one is mobility. So trajectory data, understanding why people are making certain travel decisions at what point in time of the day at a very, very aggregate level. There are some national-wide efforts. For example, the national household travel survey for a few years in the past has been a point of reference in trying to understand how people move over the course of the week. So weekday travel behavior. But one of the challenges of course is the cost of implementing this survey and the representativeness of the survey because the way you disperse these survey instruments, it used to be by placing telephone calls or polling individuals. So you really now face issues related to reaching out a widespread spectrum of the population.

The frequency at which you get data one week might be very good information, but it’s just one week out of one year. So you might really have some major issues there when you really want to get into fine grain detailed type of analytics. So now we start getting into the realm of wishes. I wish I had this data available. What frequency? Now we also started off with the issue of safety. So all the crash databases, which are depending on where you are, depending on which state you are and depending on what the practices are of collecting and reporting this information, you have a certain level of availability. As a researcher, depending on what I work on, I have access to detailed crash analytics from a reporting mechanism where I might get into the minutia, but we get detailed report information that are originated from a police report, the long form report.

So we call that database, but it takes a certain period of time to process the information for accuracy because it could happen that the crash is reported in a road, but that perhaps the police officer put as a location actually the point and space in time where his car is. So it might be parked on a parking lot and reporting that as the location wrongly of the accident. So we start getting into, we have availability of data, but there’s also some inherent bias and error and that I think it is part also of the discussion. So this is what we’re working on right now. There are new sources of data that we’re finally, I would say acquiring for which I think we can discuss a little longer down the road in our discussion, but this is what I’m working on at the moment.

Sandi Dunmyer, Arity:
Sounds great. I’m hearing a lot of recurring themes here with we need not the bias to be the data, to not be so biased. We need more of it. We need it to be more accurate and consistent and current. So I think we’ll hear a lot of that. All right, Anthony, this one’s for you. Why is it important to have multiple sources and data fusion?

Anthony Johnson, Arity:
It’s important because every data source has strengths and weaknesses, right? There are pros and cons, be it coming from mobile device, connected car plugin devices. They’re all going to have strengths and weaknesses. And I think the first part of that is really understanding what is this data and how should I use it? What are use cases that are valid to use that type of data source. Now with that said, we’re trying to understand ground truth, a full picture of what is happening on our roads and our highways, and there’s no one data source that’s going to cover that. And so that’s where the beauty of data fusion comes into place and all that is right, we’re just saying bringing in multiple data sources that have a relationship to one another, and as we start doing that, we start filling in the gaps. So data that’s coming from Arity and data that’s coming from Michelin and connected car and weather data infrastructure, you see where I’m going with this now we start getting an idea of the full picture of what’s happening on the road. And that’s as a solutions engineer. I mean, that’s a beautiful thing. I’m a data guy, so the more information the better. But yeah.

That’s where the beauty of data fusion comes into place…[We’re] bringing in multiple data sources that have a relationship to one another, and as we start doing that, we start filling in the gaps.

Sandi Dunmyer, Arity:
Erik, do you have anything you want to add to that?

Erik Dietz:
Yeah, I think just continuing where Anthony left off a little bit, understanding the context, especially of driving behavior, right? We’re seeking to improve safety on our roads. We’re seeking to improve the equitable mobility and sustainability. Understanding that context, whether it’s micro weather, it’s the infrastructure data, it’s what else was happening there. And that’s what really starts to take the data from simply lines of code to something that is actionable, something that our end customers, whether that’s research, whether it’s an infrastructure owner operator, an engineering firm, someone else who can take that and say, I need to make these decisions based on this information that I didn’t have previously, that I now have a more complete holistic understanding of the situation and what is going on. So you’re not looking at individual data sources in isolation, but you are fusing those together for a better understanding. That is then actionable. And I think that’s something that we’re very happy to work on with Arity when we talk about leveraging both LBS [location-based service] data, connected vehicle data, bringing in those additional data layers to really create that picture that creates valuable insights that we can action and improve towards our goals really on roads and mobility.

Anthony Johnson, Arity:
I’ll just add one quick thing because there’s one word that Erik said actionable, and even as a data vendor, we’ve challenged ourselves on that, right? I mean because it’s all relative, but we have a lot of data and sifting through that noise, so to speak, to really get down to what’s actionable about this. It’s almost kind of like a litmus test really to say at the end of the day, what decision can I make on this analysis or this data set that’s being presented? I think if right, if you’re not able to bring that to the finish line with an actionable decision that can be made, hey, maybe we need to retool and revisit what data sources that are involved or whatever it is we’re trying to accomplish. So

Sandi Dunmyer, Arity:
Absolutely John, Dr. Concas want to add anything?

John Hansen:
I look at the end user in this discussion we’re having, and I think you are a big part of that and that is to help define what it is, what type of information do we really need to have and there’s a tool to do our jobs, so as part of the discovery of this information, how we’re going to use it, that too needs to be part of it. I think in the past where we’ve had these watershed moments in this industry with events like data and digital integration, that piece hasn’t always been there. So part of this answer is you and the audience being a part of that answer, it’s time to come out of the stands a little bit and be on the court and play with the rest of us that are involved with this.

Sandi Dunmyer, Arity:
Okay, we’ve covered this somewhat, but I think we should do a deeper dive into any inherent biases or blind spots in the currently available data sources, things that we need to be mindful of, things we need to address. Erik, I’m going to have you start on that.

Erik Dietz:
Yeah, absolutely. I mean every data source has a bias. I think we can all agree with that and you can tell any story with a single source of data that you are compelled to find. And I think it comes back to fusing different data sources to remove some of those inherent biases. So when we talk about mobility data, the first comparison is always between connected vehicle versus LBS data, right? Mobile phone data. And I think from Michelin’s perspective, we are incredibly invested in extracting that bias. So the combination of connected vehicle data that might be very high frequency, but we also know is not equitable necessarily because connected vehicles, whether direct from the OEM or aftermarket are going to be newer, more expensive vehicles. We may not have as many of them in our rural areas. They’re going to be concentrated in our metro areas.

Combining that with LBS data, especially the type that Arity is able to provide, gives us a much more equitable perspective and coverage across really the entire population, right? When we look at the ability to have coverage in every census block group to better understand the mobility really of all rather than just a specific portion of the population who is purchasing vehicles maybe from a specific o e, having that broader perspective I think is incredibly important. And then layering that with again, the different types of data sources, whether it’s census data, whether it’s weather data, whether it’s infrastructure related data so that you have that more complete picture. I would say it’s not been something that’s been possible for very long. It’s really only in the last five, 10 years or so that this has been coming online more and more as we’re seeing, I would say more connected objects to have that very equitable perspective to pull out that bias.

Sandi Dunmyer, Arity:
To add to that, how about the different types of roads out there and not just the different types of cars but the highways versus the arterials and such? Does anyone want to speak to that?

John Hansen:
Kind of along those lines, let me first talk about when you’re putting any kind of information, especially in rural areas, there’s two questions you have to answer. What is my power source and what is my communication? Without an answer to those two things, I really can’t deploy anything of value in the rural areas. So that’s really the big challenge. The type of roads, the consistency, the weather elements, all those play into that. But at the end of the day, the starting point is really to have the communication of power identified so that this tool can be used where we want it not as a latency issue, but as a real time issue to help with what we’re doing out in that field.

Erik Dietz:
So if I can just respond to that quickly, I think it’s a great point, right? Because when we talk about power source connectivity, it’s absolutely a challenge in the marketplace. When we look at connected vehicle data, again, frequency can be very high power sources is the vehicle, which is great, but connectivity only exists when it’s in an area that it’s able to be connected because it doesn’t have local storage. One of the benefits of complimenting that with LBS data is power sources the phone and as long as you’ve got it charging, you’re good. But connectivity, even if it’s outside of cellular networks, it has local storage. And so it’s uploading after the fact. And it’s one of the greatest benefits for Michelin working with Arity is that we don’t have the same types of blind spots. When you look at rural areas, especially very remote areas, we have a pilot that’s just getting ready to go with Idaho DOT right now where we don’t have that same gaps in our coverage as a result because of that local storage on LBS, when we look all roads across Idaho or states that have significant rural areas, we don’t have those gaps.

We can see what is happening and we know that when we talk about safety issues, especially in mobility, whether it’s vulnerable road users or drivers, it is those rural areas that have some of the most dangerous hot spots with the least amount of data and the least infrastructure in place to address those. And what we’re able to do is to bring that information so that we can make those informed decisions in those areas.

Dr. Sisinnio Concas:
It is great, but I think in terms of biasness as well, we have to consider that everything so far we’ve discussed it’s a vehicular type of data and then the questions are related about the frequency, the need at which we need to have a certain level of frequency for depending on what we’re trying to achieve in our analysis or ultimately as goals. But what about representativeness? Because we’re also in a multimodal ecosystem at this point, we’re moving towards a way of moving in traffic or across regions where car, yes, it is still fundamental in the United States, but also there’s other options and in the future as we project ourselves, maybe transportation will be more of a service. So the relationship between, for example, pedestrians and vehicles requires certain type of data. So I think there is an element of biasness in there of representativeness when you want to understand conflicts between vehicles and humans. So we have to think where are the data available out there, how we can think of fusing them together to have a more complete picture so that we can reduce this bias in this.

Erik Dietz:
So I think it’s a great point, and it was a little bit what I was saying before but I might not have been perfectly clear. So when I reference LBS data, I’m talking not connected vehicle data, but I’m talking connected device, so mobile phone and that is that compliment to reduce that inherent bias that we see in connected vehicle is really to combine connected vehicle with connected device. So when we talk pedestrians or people on bicycles, on scooters, something else, we are still collecting and bringing in that data. Again, vast majority of people have mobile phones and smartphones. So when we look at having a equitable distribution of data coverage, what we are targeting in partnership with Arity is typically somewhere between a 15 and 25% coverage of all mobility users in any census block group. It’s something that we are able to achieve with that combination of LBS mobile phone data plus connected vehicle data that really allows us to provide those representative and equitable insights and information to our end customers.

Sandi Dunmyer, Arity:
Excellent. Okay. What are some key metrics around safer roads that your stakeholders expect to be improved? And do you have the tools to measure those currently, Dr. Concas,

Dr. Sisinnio Concas:
I think we have Vision Zero. Of course you want to bring the fatalities to zero, that is paramount. So that is the main goal. How do we get there and how can we use information So to proactively intervene, it could be as small as a section of your roadway. If you’re a small manager of a small section of a roadway, like as a researcher working with local transportation agencies could be an authority that might be managing a tall section of the road. Really the first thing that they have as an objective is reduce fatalities to zero and do anything that it can get you there. It could be the question becomes in terms of the resources associated to get there to process information within, didn’t touch one aspect of the vast abundance of information because yes, we are at a juncture where we are now facing ourselves with a vast abundance of information coming from sensors coming from, I’m going to use the augmented probe vehicles more than connected vehicles because we need to discern also the definition of connected vehicles in this environment here specifically at this conference, that information is out there. There’s a lot of information there is coming towards you, I say towards researchers, towards agencies. Now there is a cost of filtering through the information and also the processing of information. There are costs associated. So as you reach that goal of reducing fatalities to zero, there are there’s more information available, but there is inherent increasing the cost of getting there.

John Hansen:
And I would only add to that, that if I could use a word picture, it would look something like this table that we have our current stakeholders sitting at isn’t big enough. It’s got to be bigger. There’s a lot more stakeholders out there. There’s a lot more that we need to recognize and bring to that table to help us achieve the goal of zero fatalities. And there are a lot of stakeholders out there that are struggling with the same issue. We have got to learn to come together and work through this difficult time that we have with our fatalities as a working group together. And such areas that I’m speaking about is areas like commercial vehicle operators, fleet operators, there’s a number of other areas that just in the transportation community, we assume all of us are together, but they may not necessarily be the picture. And then add to that as we’ve heard academia and others professionals that are able to help us achieve these things, but also to speak to those areas that we need to address those things. We don’t know. We don’t know.

Sandi Dunmyer, Arity:
Okay, this is a big one. So how is driving behavior or driving events useful or relevant to enabling safer roads? Who has to start on that one? Anthony,

Anthony Johnson, Arity:
I’ll kick us off.

So traditionally we’re getting these police reports, which yes, that’s useful information. Someone was speaking to that earlier. That report has to be interpreted and determine, hey, what relevant information can be gleaned from that report and what have you. Now, as a part of that though, it is a police report. So unfortunately there’s been an accident, potentially bodily injury, property damage, et cetera. One of the ways that the data fits in terms of improving road safety is and take for example, Arity and capturing along with this trip, right, driving trip from point A to point B, we’re also capturing all of this driving behavior or what we call risky driving behavior of these events from rapid acceleration to hard braking to speeding to distracted driving, which for us we say distracted driving, we’re talking about being on the phone, unlocking locking of your phone, phone movement, et cetera.

And then also crash detection. So all of this information, all of this data, data is being reported. It is a way to be proactive to get in front of that accident of someone being injured and god forbid, loss of life to identify, hey, what are those risky intersections in the rural areas? What are those dangerous stretches of road where we are starting to see these different patterns and how can we get ahead of it? So I think that’s a great way. Well that’s one, and I’ll mention one more infrastructure. So going back to road infrastructure and identifying, hey, where do we need to put our focus and efforts in terms of improving roads from whatever aspect it may be from signage to the actual road itself, leveraging that data also, Hey, why are we seeing so many hard braking events in this particular area? So I think that’s two ways that it can be leveraged and taken advantage of.

Erik Dietz:
And maybe if I could just continue on that. So something that we collaborate along a lot with Arity is on there driving events and the information that we can pull out of that. I think one of your earlier questions is what is the data that we don’t have yet that we wish we had? And I think a lot of the co-innovation that’s happening between Michelin and Arity right now is looking at what new event types, in addition to the ones that Anthony mentioned, can we create, collaborate on and start to extract more contextual information on things like wrong way driving events, things like road departures, contextual speeding events, not just excessive speeding. So how do we continue to innovate in that space so that we’re pulling granular data that’s at the right moment, at the right time for the right use cases that then create those actionable insights.

And I think driving events is a way to do that that again reduces the overall data that’s going up into the cloud. So if you have that on the edge calculation and capture of the event, you don’t need all of that breadcrumb trip data going up into the cloud. And while we definitely have friends here from Microsoft and Amazon and others, it would be really great to have a more sustainable data model that doesn’t require those massive transfers of data, but really capturing on the edge what is important and then leveraging that for the use cases we know are so important that we need to address.

Dr. Sisinnio Concas:
Yeah, that is a great point and we’re probably getting a little bit on some of the needs in and around the events speaking still standpoint of the research, really interested in slide before and after the event instead of looking at the event as a discrete point and all the other elements and pieces of information that are attached along to these data sets, which are very, very informative, but it’s also it is due the big question, is it the infrastructure causing that event or a response to the driver to something along the roadway or is actually the behavior of the driver under say, stressful conditions. So that requires in and around the event having availability and now we’re falling into the realm of few things that have been working on for the past few years really on connected vehicle data, which is the super high frequency 10 times per second. So as a potential say, user of this novel data, one of the discussions and the cooperative discussions that might take place is like my need are on a wish wishlist, maybe 10 times per second in and around event can that be achieved? What is the cost of getting there, which is what you were pointing to.

Anthony Johnson, Arity:
Yeah, one other thing, add, sorry, yeah, that I’ll add to that because…absolutely. What types of behaviors were occurring prior to right before the crash? One of the things that we do is crash detection. So it’s algorithm that’s taking all of these. You have the accelerometer and gy, derometer in your phone, et cetera. All those measurements and G-forces, et cetera, that’s being captured. So that’s how we put the prediction out there. Hey, here’s the probability that there was a crash that occurred. We also have what’s called a collision payload. So yes, we have the event, but also what happens 60 seconds prior in 90 seconds after. So all of those measurements and things that have been captured to really dive into that data to determine, hey, what was really going on and why did that occur? We have the data to do those types of things now.

Erik Dietz:
And so that event payload that Anthony is mentioning is critical to our analysis as we go to further contextualize those events and understand what was happening 60 seconds prior around them. Is it something that they were reacting to the infrastructure? Is the micro weather data telling us that it was an extreme rain event at that time, or do we see that the solar position is such that it’s coming directly into their eyes and they couldn’t see as they were cresting a rise, right? Understanding the context leading up to that event and then the response directly afterwards for that 90 seconds afterwards, those event payloads and that we’re continuing to work on with Arity to expand what that looks like, the frequency during that event and making sure that as we create new events together, again, we’re meeting the needs of really end users like yourself.

Sandi Dunmyer, Arity:
That’s great. Something I really am noticing and really hopeful about and hopefully everybody’s excited about this, is that I’m hearing proactive, proactive after 23 years in this industry. Honestly, I really didn’t hear the word until about a year ago. So it really is exciting that that’s out there that we’re thinking about that now John wants to say something I think, okay, go ahead.

John Hansen:
I’m glad you brought that up because I hate to bring up the F word – funding – but at some point in this conversation we will have to address that, but this is a total departure. If you look at it from where we have been in the past around planning versus where we are with this discussion in the past we looked at, okay, how can I make improvements on the highway to make them safer for the road? I can’t build a new highway, so I have to use infrastructural changes to do that. That costs money. Now we’re talking about predictability, knowing where accidents might occur, how we may be able to react would be proactive is a real new way of looking at this whole picture. And I hope you’re all capturing that.

Erik Dietz:
I think it’s a great point. And to be fair, I think the reason it hasn’t been discussed a whole lot in prior years is it wasn’t possible, right? You didn’t have the technology innovations, the services that we’re now bringing to market that allowed the machine learning and the AI to start predicting what is going to happen, right? You had to be reliant on FARS data and crash data to tell you several years after the fact what happened here and that analysis. Whereas now we can look at the trends of driving behavior in an area under a given context and say, we are seeing an increase in near misses, an increase in speed, an increase in these dangerous areas, and we can say this is indicative of the exact behavior that led to a fatal crash before. And we can tell you there’s going to be a fatal crash in this area with this amount of likelihood within a certain time horizon.

Sandi Dunmyer, Arity:
Great. Thanks guys. Hey, I want to try to just give a couple more questions in before I want to make sure we have time for questions because this looks like a very inquisitive group out there. So what’s the most challenging part of introducing newer data sets into transportation planning and models? Who wants to start with that?

Dr. Sisinnio Concas:
Yeah, definitely. I would say that the challenging part is to understand, actually make, I would say try to understand whoever you are communicating to the benefit of even this information available. As I’m going to put into context, so I work for a research entity, so we do a lot of grant funded research for public transportation agencies, could be a local agency, a state agency, or at a federal level. The challenge is in communicating what are the benefits coming from using these new novel data sources because they’re just new. So when they’re on the lookout trying to reach their goal, for example, goal of vision zero, even zero fatalities, they know what ultimately the goal is, is just they’re presenting with new opportunities and new costs and probably one of the biggest challenges to make transpire what the benefits are associated with these new available data sources.

Sandi Dunmyer, Arity:
Anyone want to add anything? Go ahead, Erik.

Erik Dietz:
Yeah, no, I think some of the challenges to accepting new innovations is what we’ve seen in this industry over the last 10 years. There have been a lot of claims and I think there’s a lack of education and something that we talked about earlier is having an understanding of what is and what is not possible. And when we talk about innovations and really what we’re all here doing at ITS is bringing those innovations to market, but we need to make certain that the intent and the capability of these innovations are well understood as well as the companies who are bringing them to market. Obviously we have a lot of entrepreneurs here, we have a lot of startups here. We heard this morning at the city DOT roundtable was their concern around the resilience and viability of many of these startups. So if you start working with a certain technology, are they going to be around for the entire five-year contract or 10 year contract that you need them around to actually realize some of that change? And it’s something that Michelin really hopes to take a large part in this because again, we’ve been a mobility company for 130 years. We intend to be a mobility company for at least another 130 years. And as we partner with Arity and Allstate group, we have that faith, that confidence that we will be around bringing valuable offers to market, both internally developed as well as making investments in the ITS community overall to make sure that those offers that are creating value are staying.

Sandi Dunmyer, Arity:
That’s great. I think hopefully some of you are thinking, okay, now I understand the benefits of large scale data. I know we have more funding than ever before. We haven’t talked about funding and that can be questions, but we all know there’s more funding than ever before, but I don’t have the manpower or the extra resources to implement this. So what advice can you offer strapped agencies that want to utilize this data and how can they get started? Anyone can answer this one.

John Hansen:
I think we touched on it a little bit earlier. They have to be a part of the evaluation and part of the answer, but you bring up a really good point. They’re doing a lot more with less. We’re all aware of that. And the technology that is a part of this is something that a lot of agencies do not have that understanding in that if you will, tool set in their toolbox at their organization, which presents for us the challenge of not only assisting with the tools, but like a dashboard approach to help them utilize it. There may be things we can do with technicians to certify ’em in this area. I don’t know what the future holds, but I do know it holds that we have to consider the support side and the maintenance side of what all this represents.

Erik Dietz:
So maybe if I can just because I see Omar here from Ottawa, who’s one of our customers we’re launching a pilot with right now. I think when we talk about helping them leverage and resource and workforce development, the conversations that we had with Ottawa to clearly define what the need was for your city to understand is a variety of projects for us to look at before and after scenarios to help understand a safety diagnosis across your network to identify where some of those areas of concern are for vulnerable road users, having that open dialogue and really co-create the solution, what are you looking for? And I think those open conversations are something that there’s a great opportunity to have more of. And I think too often private sectors very focused on just selling and I think we need to realize we need to pause, we need to ask questions, we need to seek to understand so that we can collectively bring value together to the marketplace to solve some of these issues.

Sandi Dunmyer, Arity:
Any good relationship has lots of listening. Go ahead. Anthony, did you want to say something?

Anthony Johnson, Arity:
Yeah, no, I agree with everything that was just shared, and John touched on it earlier, which I can’t stress enough of the consumers of the data, the public entities, agencies to you say, come out of the stands onto the field. I mean, it will take a community effort and have open honest dialogues. I think we are well aware that we don’t have all the answers. We know we, we are sitting on a treasure trove of data, but in terms of all of the different applications, and again going back to is it actionable, we need that feedback to help us navigate which way do we need to go in terms of limited resources. I think that that takes open and honest conversation also because yes, we can sell you data, but in many Arity as well as I’m sure other vendors out there, there are opportunities for service engagements who say, Hey, at the end of the day, this is the analytic that I need on this data. Can you guys provide that? And I would also say, which I’ve carried this with me my whole career is be brilliant on the basics. We recognize that it’s new, that the very first thing that we do doesn’t have to be right this massive undertaking in all of these analytics. Maybe it’s some low hanging fruit, right? Let’s baby-step it, be extremely successful at that and then continue to build on top of that in terms of introducing other analytics data sources and things of that nature at the same time, that’s building the confidence too. Maybe I haven’t worked with this data source before.

Sandi Dunmyer, Arity:
That’s great. Okay, I think we’ll close in time to take a couple questions here, but we collectively really hope that this discussion was the start or the continuation of industry efforts in creating safer roads, making use of data that’s available. Does anyone have a question they’d like to ask somebody on the panel? Hey, we have a gentleman. Oh, sorry, Janet, would you mind passing him the…Thank you.

Erik Dietz:
This is really a great discussion, but no one spoke about cybersecurity and cyber risk to really make the transportation safer. What’s your take on that one?

Sandi Dunmyer, Arity:
Could you guys hear? I couldn’t hear very well.

Erik Dietz:
So if I can just reiterate what I think I heard asking about cyber threats and cybersecurity and how that interacts with the data that we’re collecting and…

Anthony Johnson, Arity:
Yeah. Well, I’ll take a swing at it. So it is been a journey. I’m speaking in terms of Arity, right? We recognize that the data that we capture is very, very sensitive. We’re talking about trip data, driving individuals going from point A to point B – and A to B could be a residence, it could be a healthcare facility, place of worship where I drop my kids off at school. And so de-identification anonymization is paramount. And that’s something that’s really the first step in even looking at. So trimming off, for example, the first quarter of a mile and the last quarter of a mile of a trip. Now in the last 12 months what’s happened, we’ve taken it a step further beyond that to where everything that we deliver is in aggregate. So we do not sell individual trip data. Everything has been rolled up either to a geography, be it zip code or a county or state. We’re rolling out data on the trip date and things we’ve been talking about match to the road segment, which is coming out later this year.

Erik Dietz:
And I think if I can add to that, I think it’s an incredibly important question, right? Because when we talk about data Arity, Michelin, we are stewards of that data and we have data that’s very actionable and we want to make sure to the best of our abilities and we continually push the boundaries of what the best of our abilities is to be a good steward of that data. Certainly as a French company, we adhere to GDPR and if there’s ever a trade off with CCPA, we pick whatever is more stringent there. But I would say more often than not, mobility intelligence is pushing the boundaries of what we should be doing far and above what is required by law. And I think one of the reasons that Michelin values the relationship with Arity and Allstate group so much is we are in the business to take care of our people. And I think that privacy, that integrity, that ownership is extremely important to both of our brands. And the one thing that we always kind of laugh about is I can’t put our group CEO on “60 Minutes.” So it’s something that we could never have something like that. So we go above and beyond to make sure that we’re not ending up on the news saying that we have mishandled personal information or people’s individual locations, we can reach actionable insights without that.

Sandi Dunmyer, Arity:
Yes. Mr. Blake – would someone, pass him the microphone.

Audience member:
So yesterday morning in the big data session – is that better? Yep. So yesterday morning in the big data session, they talked a little bit about the difference between real-time data and archive data, and somebody made the comment, well, we only really care about real-time data. And I think that’s wrong. I think we need both because there are different use cases for both. So we need to understand, for example, I have a client, like my manager comes to my office or to the TMC all the time says, okay, how’s traffic today? I don’t know, right? Because you need that real-time data to be able to say what’s going on in the past five minutes versus what’s been happening over the past 30 days. So talk to me a little bit about how you see the type of data that you can provide serving both of those functions or not.

Erik Dietz:
So I can start with that. And then Anthony, if you want to add on, I think you’re absolutely right that there is a place for both. And I think also the definitions here are really important because people will say real time, but maybe what they actually mean is a three second latency or maybe it’s a five minute latency, maybe it’s a 24 hour latency, and they still consider that real time. So clearly identifying based on the use cases in the customer, what is real time and what is not, and then what is historic, right? Because there are many people when they talk about historic, it’s three years old, it’s FARS data. So there’s a huge continuum there for the latency or the freshness of that data. I think when we talk safety related use cases and planning, we’re typically looking at data and insights that are as fresh as two to four weeks hold.

That’s aggregating at a level that we can actually pull out some actionable insights that is cleaned. It is contextualized. We’re transforming and layering and fusing that data in the ways that we described earlier. And we talk about traffic management, certainly I think sub five minute latency is really important, but what you get as a compromise with that is the quality, right? Is it, is it as accurate? Is it contextualized? Probably not. So you’re using very specific sources to provide an understanding of that real-time traffic or near real-time traffic. Those are different use cases with different operators leveraging that for, again, very different purposes.

Anthony Johnson, Arity:
Just to quickly add to that, yeah, I totally agree. You do need both. It depends on the use case. Arity is a data vendor and what we’ve learned is that’s all about discovery and asking the right questions because what real time may mean to me may mean something totally different to another individual. So we have to, again, going back to that actionable, what are you trying to solve for? Because I’ve had multiple conversations where we get on the Zoom call, we need real time, we need real-time, Dave knows. But as we start talking and really start flushing out what the use case is not so fast, it’s actually the historical data will answer the questions that you have. So I think that’s what it boils down to.

John Hansen:
However, [laughs] if you go around a room right now, there’s going to be a number of vendors here that are talking about the ability of changing dynamically, the traffic signals based on conditions, volume, those types of things. Look, if we’re going to do that, we have to have real time information, not just a signal that is at that location that is part of the equation. It’s all the signals that comprise a corridor that are part of that equation. So real-time data is a blessing that we’re looking at to be able to achieve that. So there’s a real world example of something that all of us are striving to do today that these tools are going to help us achieve.

Sandi Dunmyer, Arity:
Thank you. And I know there’s a couple more questions, but I’m going to ask the experts to stick around for a couple minutes. We are out of time, so just want to say thank you guys so much for joining us. Thank you, experts. It was really an intriguing conversation. I know we could talk for hours more, but I hope that people are leaving with the three takeaways that we hoped regarding the where have we been and where are we going with data, or at least get you started thinking more about it, what the data means to us regarding safer road mandates and also how do you implement these technologies. So again, if you want to talk to us more, we’re straight back this aisle. Michelin probably kind of hard to miss with the Michelin man there. And John and Dr. Concas will stick around as well for a couple minutes. Thank you all so much.

John Hansen:
Thank you.