Using the OpenPATH Trip Tracking and Survey Tool to Collect and Analyze Travel Data (Text Version)

This is a text version of the video for Using the OpenPATH Trip Tracking and Survey Tool to Collect and Analyze Travel Data presented on April 19, 2022.

Lauren Reichelt: Hi, everyone. We are recording. Hi, everyone. Thanks for joining us in this first webinar in this new Coffee with a Researcher EEMS webinar series. This series is part of a larger effort to better connect the Department of Energy's EEMS research with Clean Cities coalitions and technology integration.

We want to open the lines of communication between coalitions, their stakeholders, and researchers working on EEMS efforts at DOE. These sessions will highlight available EEMS tools and insights, and help coordinators identify the local and regional partners, and the projects that might benefit most from these capabilities. We want these sessions to be conversational, and provide an opportunity for EEMS researchers to ask for input from coalitions, and for coalitions to engage with EEMS. So, Clean Cities coalitions, as you know, are in a unique position to describe the mobility priorities and challenges in your own communities, and researchers can use that to enhance EEMS research. We want to emphasize that while the featured researcher will start by sharing what they're working on, we really want to open up to more of a conversation afterwards.

So, Shankari, who we'll be hearing from today, is really excited to talk to you and find ways to collaborate. Please, collect questions throughout the presentation, ideas for how this might apply to your work, or even how this work could be improved in future iterations to be more applicable to community needs. And, as Cass had mentioned, please, feel free to post any questions in the chat and then, we'll have a starting point, a jumping off point for when the initial presentation concludes. In addition to this series, we are doing a number of other things to connect Clean Cities coalitions with EEMS research. So, we've sent some Clean Cities coordinators to the annual merit review to gather insights and to begin to understand the opportunities and challenges that coalitions face around EEMS projects.

We're also developing educational materials, including an online course on EEMS topics for coalition staff that'll be on Clean Cities University, and a brochure about EEMS that coalitions can provide to local stakeholders. And we've also started an EEMS Clean Cities working group of coalition coordinators to provide us with regular feedback, to review education and communications materials, and programming like this, and help us continue to improve. And, with that background, I would like to kick off today's webinar. Energy efficiency through behavior change is a key leg of the transportation decarbonization stool. In this session, you're gonna learn about an easily deployable tool that communities can use to develop or evaluate programs that are focused on behavior-based transportation shifts and efficiencies.

And I'm happy to introduce today's EEMS researcher, K. Shankari.

K. Shankari: Hi, everybody. My name is K. Shankari. I'll be talking to you about OpenPATH – NREL's open platform for agile trip heuristics.

I'd like to start with an overview of the twisty journey that has brought me here. I worked with computers for my entire professional life. I did my bachelor's in computer engineering from the University of Mumbai, and my master's in computer science from the University of California Santa Cruz. I then spent 12 years working on distributed systems in the PEC industry. I was always interested in sustainable mobility.

I used to take my kids to preschool on an electric bike, for example. So, when I wanted to do my small part towards tackling climate change, I went to UC Berkley to get a Ph.D. in computer science with a designated emphasis in global metropolitan studies. I then joined NREL as the director's fellow to continue my thesis work on the behavioral aspects of transportation decarbonization.

Last spring, a group of us at NREL submitted OpenPATH – then called e-Mission – to the Energy I-CORPS program and were fortunate enough to be selected. The I-CORPS program is a DOE funded intensive program that provides researchers with expert advice to take their technology out of the lab. We had to conduct a minimum of 75 customer discovery interviews with various stakeholders across the ecosystem. We ended up conducting 78. We used those interviews to determine who cares about this tool and wants to deploy it.

Through this process, we were able to identify that our key stakeholders were public agency program managers. Such managers frequently need to evaluate projects or programs to incentivize clean transportation. Consider the case of a public agency that is looking for a tool to evaluate an employer trip reduction program. Big data solutions are not linked to participants, so, we cannot assess the underlying biases or use them to detect individual behavior change. Town hall meetings are typically attended by advocates – both GIMBYS and NIMBYS, who show up at every meeting to argue.

Web-only surveys don't capture travel complexity if they're short or are so long that nobody wants to complete them. So, what is the solution? That's where OpenPATH comes in.

OpenPATH is an open source extensible platform for instrumenting human mobility. It consists of a smart phone app, a server, and an analysis pipeline for generating a linked end to end multimodal travel diary. The platform displays the diary to the user, along with configurable buttons for semantic information. Detailed information is made available to program administrators through a private dashboard, while advocate statistics are published publicly. The generated data is similar to classic travel surveys and can be stored in NREL's transportation secure data center.

I'd like to elaborate on OpenPATH functionality through a concrete use case – the Colorado Energy Office e-Bike Program. Through this program, the Colorado Energy Office provided e-bikes to low-income households. During the mini-pilot last fall, we used a custom version of OpenPATH used CanBikeCo to collect data for evaluation through tracking all participant travel and asking participants to label both the travel mode and the replace mode. We were able to determine the difference in energy intensity for each trip. We combined that intensity with the trip length for trip level impact, and then, added it up to evaluate the program as a whole.

We found that e-bike trips typically replaced car, walk, and bike trips. The car trip replacement resulted in emission savings, while the walk and bike replacements actually resulted in a small amount of increased emissions. However, the e-bikes are so efficient, that the overall impact was strongly positive. Note that the walk and bike replacement actually increased participant productivity, think MEP, so, the program actually balanced equity and sustainability goals.

Using OpenPATH was critical to this evaluation. Without OpenPATH, the CEO would have most likely used web surveys – such as Survey Monkey or Google Forms – to evaluate their program. In fact, the mini pilot actually included a weekly survey for participants. The survey included open-ended qualitative questions, but it also asked questions about e-bike usage and mode replace in the past week. Unfortunately, while, it is somewhat feasible for participants to estimate the count of their trips in the past week, it is extremely difficult to estimate distance.

So, if you wanted to estimate energy emissions impact by multiplying the difference in energy intensity by the trip length, we would not that trip length. In contrast, OpenPATH maintains an auto-generated record of all trips. So, this allows us to get trip by trip lengths and calculate this impact. The side-by-side comparison also reveal that many users did not complete the survey, and among those that did, there was significant over-reporting of e-bike trips. Such over-reporting biases are well documented in the travel survey literature.

In the absence of OpenPATH, the classic solution to avoid reporting biases is to use passive data – typically, through vendors such as street light data. These vendors purchase data from location brokers who, in turn, aggregate their data from a multitude of smartphone apps that collect data opportunistically. The vendors are able to analyze the data and generate linked-accounts or origin destination flows. However, these trips are not typically linked to each other or to a user. This makes it hard to assign more complex labels to the data.

For example, returning to our energy outcome, we need to know the mode and the replace mode for each trip to detect the difference in energy emissions intensity. But it's hard to distinguish between regular bikes and e-bikes, and, for that matter, it would be e-bikes and automobiles on city streets using sensor data alone. We're not aware of any passive data vendor who even attempts to distinguish between regular bikes and e-bikes at this time. The replace mode is also very hard to input automatically, since it depends on participant thought processes. There are also other questions that add meaning to the automatically collected data, which will be hard to infer without round trip.

Now that we've seen why OpenPATH is useful, let's dig a little deeper into the components with some live demos.

Participants can install the OpenPATH related app through the stores, just like they would install any other app. The app will typically have a summary of what the program is designed to do, and then, a detailed explanation of the data that we collect, how we collect it, where we store it, how we associate it, and what we're going to do with that data. After they have read through that and accepted it, they can turn on the various sensors that the smartphone app collects in order to build the travel diary. So, the thing to remember is that everything here should be green, and if it is, then the app will work properly. We then have the ability to log in with a token, which is the short phrase that we give users who are participating in the program for the eBike program.

They were actually given a token along with the e-bike and so, they use that to log in. We don't need to then know their e-mail or their phone number or anything else and can keep that PII out of our systems. This is a demo so, I'm logging in with a token that has my name in it and basically represents my data. So, as you can see, this is basically all of my travel for the past few days. I was very busy yesterday and I was even busier on Sunday.

We have a summary of how long a trip took. We have an estimate of the distance, and we have the start and end point of the trip. If we actually go and look at the diary view, you can see, on a map, where this was. So, this was me going from my house to drop off a camera that we had rented, and this is me coming back. You will notice that some of these labels are yellow.

That's because we built some functionality to automatically guess the labels. I'll talk about that in a little bit in greater detail towards the end of the presentation. Once we have all of this data, we can actually estimate the individual carbon footprint for any particular user. You can see here that I have, you know, a breakup of my travel across various modes. I haven't labeled all of my trips, which is why this is a range, but if I had labeled more of my trips, then this would actually show a single value.

We compare that footprint to the average for the group and how much you've saved, but we also compare it to what the weekly goals for per capita carbon footprint would be for meeting the US 2030 and 2050 goals. There's a profile where – which is not important in this point. So, once we have all of this data at the individual level, we can actually aggregate it to show how the program is doing as a whole. So, you can see here, these are the various locations where the Colorado Energy Office program is active right now. So, after the mini pilot, which was in Denver, the program was expanded to multiple locations across Colorado so, you can actually compare how the programs are doing in various locations in terms of mode share and so on.

So, 4Core, which is in the Four Corners area, is doing extremely well, but that's because I think they have a few people who don't have cars and cannot drive cars. The other programs are doing reasonably well – you know, about 30 percent e-bike share, which is actually really good from a mode shift perspective. Vail, which is up in the mountains, probably needs to warm up a little bit more before people will actually use their e-bikes. You can actually compare this over time. So, for example, I can pull up from March Smart Commute and compare how they were doing.

So, in March, Smart Commute – there weren't a lot of people riding e-bikes, but they really improved in April as the weather improves. So, you can have this sort of longitudinal picture over time. And, of course, you can also pick any month and metric. So, the energy and emission impact metrics that we saw in the slide show – you could just add that.

Whoa. This was all because of bus. Let me pick a different one. Maybe I'll do community cycles for this one. So, you can see that again, here, there was a lot of replacement of regular bike. There was actually a significant replacement of bus and not that much replacement of walk, again, because this is January/February and I guess it was cold out in Boulder, but there was a lot of bus and car replacement. And I want to highlight that the bus is actually – riding an e-bike is actually more efficient than taking the bus not just in terms of time, but also in terms of energy efficiency, because the bus occupancy's not very high outside or urban cores.

So, that's sort of the view that – this is public. There is no information in here that is specific to any individual so, this is just – you can go to this website today, now, and check it out and explore it and see how the programs are doing. For the program administrators, we actually have a separate dashboard that has a lot more detail. So, this actually lists out all of the participants – this is our staging dashboard. So, it lists out all of the participants, what phones they have, how many trips they have, what percentage they've labeled.

It will let you see sort of individual trips one by one. This one, for this user, was on the 15th and then, 12th and so on and so forth. These are the labels that people have provided for their trips. You can actually go up here and do some rough histograms. So, there's some outliers here, but a vast majority of the trips are short, which is what you would expect given the census data.

You can also filter based on various things here. So, we could say we only want to look at eBike trips here so that we can compare it to drive alone, for example. So, these are only bike rides – and you can see they're a lot shorter. And then, finally, we actually have a map representation in here where, again, not gonna make this public, but we basically have, by default, the origin destination locations. These are actually color coded by the trip mode, and if you click on "Get Locations and Trajectories" it will actually show you the full trajectory of the trip when it finishes loading.

So, that's roughly what we have for the program admins right now so, using this, some of the programs have been able to ask their participants, "Hey, why are you guys not riding? What's your problem and what can we help you with?" And they were actually able to purchase winter riding equipment for some of their participants in order to allow them to utilize the e-bikes better. So, I'm gonna pause here to see if there are any questions with this sort of first part of the presentation 'cause I know that was a lot of information to take in, and then, I can move on to the second part of the presentation.

Lauren Reichelt: Shankari, I think we have a few questions. Sonja has asked, in the chat, "So, how does the tool determine the mode of travel? Is it based off the traveler's speed or is that put in manually?"

Shankari: So, the answers is – there's both. So, we have a set of algorithms that determines the mode and I'm happy to go in excruciating detail about this, because this was my thesis and so, it's – the algorithms are actually publicly available. You're welcome to look into them as well. We don't use only speed. We actually use the motion activity sensor from the phone, and then, do some smoothing around it to determine sort of contiguous sections of a trip.

And then, we use speed and a combination of a GIS integration to do the five basic modes like walk, bike, bus, car, and train. But we also allow people to put in labels. So, with the sensor data alone, it's very hard to distinguish between say, car and carpool, 'cause in both cases, and you're going at a car speed. So, we do allow people to put in labels as well. Programs, to some extent, can think about what they want to incentivize.

The CU program – at least initially – they wanted to incentivize labeling. Over the summer, we're thinking maybe of falling back to something where we ask people to label some of the time, and then, we fall back to the automatically sensed modes if they haven't labeled, but I'll get to that towards the end of the presentation.

Lauren Reichelt: Can it tell between an e-bike, a regular bike, and a scooter, for example, or is that something, again, where you need that manual labeling?

Shankari: At this point, the automated algorithms do not distinguish between bike, e-bike, and eScooter. I'm not gonna commit to this over the summer, but over the course of next year – or maybe next summer – we will have enough e-bike data that we can try to distinguish between bike and e-bike. If we had enough eScooter data, we can try to do that as well. So, I mean, if using machine learning – for those of you who are not familiar with machine learning algorithms – typically it works better if you have at least a reasonable amount of label data. You need to know something for what you know the answer is so that you can learn for it, and you can infer the things for which you don't have answers.

And so, we need that sort of massive data before we can do automated answer for things like that. I have ideas around how to tackle carpooling, too, but I'll leave the for the future web slide.

Lauren Reichelt: So, Miriam, to the point of whether these features are free for Clean Cities coalitions – Shankari's gonna get to that a little later in the presentation about how Clean Cities coalitions can access this tool, but your second question – "Can this tool be used for any vehicle or only for e-bikes? Could then, for example, use this took for EV drivers?

Shankari: Yes. So, the answer is that you can use it for anything. And, in fact, I'm actively looking for people who are willing to partner with us to explore behaviors related to other modes as well. I mean, e-bikes are great. I rode an e-bike with my kids.

But the real power of OpenPATH is it has this comprehensive view of travel behavior so, you can actually look at things like, "Are people actually driving the EVs that they buy and how much are they driving it? And does it actually replace their gasoline powered cars and so on and so forth.

Lauren Reichelt: And then, this is a really great question – something that I've heard a couple of times. What concerns, if any, have you heard or have you had to address in regards to privacy? How is privacy addressed? And I don't remember if that's part of the later half of the presentation.

Shankari: I don't think I have an explicit slide that calls that out. So, the way that we address privacy is by having an open source, end to end system that does not involve any third parties. So, the data comes directly – I wrote the app. I wrote the server. I wrote the analysis pipeline, too, and the data goes directly from an app that people install with consent into a system that we host.

So, there isn't as much concern about the data being resold in ways that the user cannot control because we're saying, as part of that initial consent document, we say, "We will never sell the data" and so on and so forth. Concretely, in terms of the e-bike program, I think there were maybe three people who were not comfortable with installing an app like this on their phone, and I think that they committed to track manually using a spreadsheet or something like that instead, but out of the other like, hundred or so people did not have concerns, as long as we made it really clear that we're not collecting any of their PII. We don't know their address. We don't know their name. We don't know their phone number.

We don't know their e-mail address. We will only use the data for evaluating the program and for further research. We will never sell the data, and we control the entire pipeline of how the data gets to us.

Lauren Reichelt: Thank you. We have one more question for this break, and then, we can go into the second half. So, how does the running of this app impact a phone battery?

Shankari: I'm glad you asked that question. So, one of the big aspects of this that took me so long to do as part of my thesis was to make sure that the battery drain was low. The sensing is automatically turned off when people stop moving and is automatically turned on again when people start moving. This can make, actually, the start of the trip a little bit fuzzy, because we – it takes some time to detect that people have started moving and then, start the really fine detail tracking, but I felt like that was an acceptable tradeoff for having low battery drain. If you look at the FAQ on the CanBikeCo website, I actually have the numbers that I got out of my thesis, but basically, the battery drain is proportional to the amount that you travel.

And I've tried it – I've traveled – I've done actual experiments in which I carried multiple phones – one with the app installed and tracking in the other one with the app installed, but with tracking turned off – so that we could compare exactly what the power drain was. And for anything up to about two to three hours of travel a day, the battery drain is under five percent. If you travel something closer to six hours a day or so, then, it can get – I forgot the actual number, but it gets to like, 20 to 30 percent, I think. There are ways around that as well. There are some advanced settings that people can set to collect the data less frequently so that the battery drain is lower, but we haven't had to use that yet, because there are a few people who travel six hours a day.

Lauren Reichelt: Great. Thank you, Shankari. And after the next part of the presentation, if any of you had follow-up questions – and, of course, additional questions – I encourage you to actually show your video and have a conversation here.

Shankari: Cool. So, now that I've shown you what the tool can do, what NREL – maybe we can talk a little bit more about how we hope to make this tool available to you and to be deployed in the world to help us understand these questions around travel behavior. So, we'd like to invite you to co-design an NREL-hosted instance of this tool. OpenPATH is open source, which means that all the code for the platform – the code for the app, the code for the server, the code for the analysis – is on the web. It's on this code-sharing website called GitHub, which you may have heard of.

So, people who are technically savvy can actually go and make a copy of this and install it and build their own apps. And I'll talk a little bit more about that later. But we want to do is to make it easy for program administrators to access the tool and so, we want to host an instance at NREL. We will not recruit participants. That's not our skill set, and it's also not something that I think would work really well because we're not part of the communities that are trying to understand this travel behavior.

So, the Clean Cities or communities that want to understand their own travel behavior will recruit their own participants. The public dashboard will allow the community to view their tax dollars at work, while the deployer dashboard will allow program managers to manage the program in real time. And our thought is that we would actually have multiple of these little enclaves for each study that wants to use it. Maybe there's somebody in California who wants to use it and somebody in Colorado who wants to use it. Their data, during the course of the program, will be separated from all the other studies.

We will allow some limited customization of the app to meet the actual program goals, and people will only have access to their enclave. So, the public dashboard will be for your enclave. The deployed dashboard will be for your enclave. And finally, we will archive the data in the transportation secure data center for long-term analysis.

So, as a concrete example, consider the case of a city which wants to attract residents to downtown areas without building hulking parking garages. The energy efficient approach would be to incentivize alternate modes of transportation, but downtown merchants are concerned that this will divert shoppers to big box stores with plenty of parking. The city could choose to ask its residents to install OpenPATH. They could partner with retailers to provide incentives for public and active transportation. This would be similar to existing parking validation programs, but with an energy efficient twist.

Program admins can monitor recruitment and retention through the deployer dashboard. The public dashboard can make aggregate metrics publicly available, providing transparency into the impact of incentives. It could also provide comparisons with other downtowns that are also using OpenPATH for their programs. Participants can receive customized surveys asking counter factual questions. For example – what might incentivize drivers to ride their bikes instead?

And finally, long-term archival in the TSDC can also enable more sophisticated analyses that can influence land use and urban policy planning. This is an expensive and ambitious vision, but we hope it gives you a taste of the opportunities that this rich data collection can enable.

But OpenPATH is not just for bikes. Its holistic picture of travel patterns means that it can be used to evaluate behavior changes for automobile modes as well. As another example, consider the case of a state which wants to reduce highway congestion and greenhouse gas emissions without expanding highway capacity. The energy efficient approach is to incentivize carpooling and public transit, which can reduce the number of vehicles on the road while continuing to transport the same number of people. But it is often challenging to find carpool partners – typically in low-density areas.

States typically address this by creating park and ride lots, which serve as meeting places for residents from multiple areas to congregate. Commuters can meet up with their carpool at the lots or transport a commute-oriented express transit. However, the emission impacts of these lots are poorly understood. For a more principled evaluation, the state could choose to ask its residents to install OpenPATH. They could provide access to prime parking spots to commuters who use the app.

They can ask commuters to mark their parking location in the app and get an estimate of parking utilization. Since the app supports a comprehensive view of travel behavior, they can determine the mode and distance of the trip after parking, which allows a high-quality estimate of the emissions impact. As importantly, they can determine where commuters come from. So, if the lot fills up early, they could provide alternate mechanisms – local shuttles, e-bikes – to access the express commuter services. The deployer dashboard, with its Excel export and trip visualization capabilities, can allow administrators to evaluate parking utilization close to real time.

And again, the public dashboard can make aggregate metrics publicly available, compare across lots to see which ones have greater utilization and greater transfer to an energy efficient mode. Participants can also receive customized surveys asking counter factual questions around why they carpool on some days, not on others, and again, long-term archival in the TSDC can enable more sophisticated analyses for long-term planning.

A common euphemism in urban planning is that all land use is local. We fully anticipate that different programs and studies will need to tailor their data collection to their specific geography and climate. We plan to support this by offering basic customization through a standard MOU that the data collector would execute through NREL. Most sophisticated customizations would require hiring an app developer – although we do request that the resulting code be contributed back to the open source project. An analogy that may help explain these levels of customization is that of event invitations.

Websites, such as Evite, allow event organizers to create invitations by selecting from a set of predefined themes. The website is clearly hosted at Evite, and the organization is typically limited to selecting a theme, uploading a photo, and entering event information. They're very easy to create and are free to the organizer. But if an organizer wants to check Covid vaccination status as part of the RSVP or offer a live stream, they cannot use Evite. Instead, they must create a custom website, potentially by hiring a web developer.

The resulting website can have a memorable URL and can include sophisticated design elements such as a hero image and animations. But these invitations can take longer to create and can be quite expensive.

NREL plans to provide an Evite-like solution where the data's collected via an NREL app, stored on NREL servers, and automatically archived in the TSDC. The data collection will be enabled through a standard MOU with NREL to avoid custom contracting overhead. We will support basic customizations. Some examples could be colors, logos, the onboarding survey, potential incentives, and so on. The server will be reachable via an URL and include a custom description of the project.

OpenPATH is open source so, if your deployer wants a more sophisticated experience, they can hire an external app developer to create a custom app based on OpenPATH and connected to their own server. We have seen existing examples of this approach in custom apps developed by Transwave, FabMob, and the University of New South Wales. They're all outside the US so, they don't want to use anything that is US hosted, but they can just take the code and build their own one that is hosted in their own country. This data does not have to be contributed to the TSDC, although we hope that US deployers will continue to do so for the greater good.

Finally, there are people who are interested in quantifying their carbon emissions, even without being recruited into a study or program. For example, we published an initial version of the app for NREL internal use only. This was very clear from the very first screen shot in the app listing, but apparently, a non-NREL user not just installed the app, but felt motivated to leave a one-star review because it was not available for public use. We would like to give such users and other stakeholders such as climate action groups – for example, – an opportunity to use the app and contribute their data for social good, however, NREL does not have the capacity to sign MOUs with each interested citizen. We have discussed supporting such users by creating an open enrollment study that is always active.

As part of the study, we could actually offer users a choice of themes. So, this is the dashboard that I just showed you – in this case, before I took this screen shot, I actually labeled most of my trips so, there's actually a fairly good number here, but I worked with some students when I was at Berkley, and they actually built this polar bear gamification where, when you sign up for the app, you get a polar bear, and depending on how you travel, your polar bear's happy or it's sad. Do people want gamification, or do they want a dashboard? Do they like data or do they like emotion?

We don't actually know very well. But if we had an open enrollment study like this and people just sort of sign up, we could allow them to choose between – this is actually a dashboard that somebody in Germany built and contributed back. That's why it says, "Your German average". So, this would have the added benefit. We would be able to determine which visual representations are most engaging so that the programs, when we choose which of these they want to do for their program, can have some data to help make that decision.

We're also working on features to improve the usability of the app and solve interesting research problems at the same time. We're implementing a label assist feature – so, I think I talked about this during the demo – which guesses the labels based on prior labels for similar trips. So, you know, if you drive to work every day, after three times, we see this is the start and this is the end. We'd be like, "Okay. Maybe you are – this is a work trip so, we can fill in the purpose, and maybe we can fill in the mode as well."

We implemented an initial version of this last summer, and I have an army of interns to improve this in the coming summer and to work on the Count Every Trip project, which I'll talk about soon. We also implemented a status screen, which allows users to navigate the increasingly complex permission settings required for proper operation of the app. Even with label assist, labeling for months can be burdensome so, we're beginning to incorporate uncertainty into the metrics. So, that's why there's a range of values here – because there are some unlabeled trips here.

This will allow us to combine user inputs, label assist values and the sensor-based inferences to provide appropriate error bars for the metrics. Participants who want to shrink the error bars and get one number can choose to label their trips.

That's it from me. Do you have any questions?

Lauren Reichelt: Maggie, go ahead and un-mute and turn your video on. Thank you.

Maggie: Hello. Thank you for the presentation. This is really cool. We have a project that we're participating in that hasn't quite launched yet, but it's gonna be an EV car-sharing project, and I'd be interested in using OpenPATH if there's a way for either the car-sharing members or the car itself to have OpenPATH built in. And then, I have a couple of other ideas about external groups who may want to use this for their city eBike programs, but that's the main thing.

Do you think that use of this in a car-sharing program, to understand, for one thing, just car sharing and its energy impact, and for EV car sharing, same thing, but we'd assume greater energy reduction.

Shankari: Yeah. I would be really interested in exploring that with you. So, OpenPATH is not really designed to be installed in only one vehicle, because then, you can't get that comprehensive picture of one person's travel. It's really designed to build a travel diary for individual people that span all modes. So, you can place the car sharing in the context of all the other travel.

If someone car shares once a week, is it because they're working from home and so, their one trip a week to the grocery store is on the car share? Or is it that the rest of the time, they're driving around in their pickup truck and, once a week, they use the car share for I don't know what reason, but maybe they're going a little bit farther and they're worried about gas prices. I don't know. But you can really put that in the sort of a context of their overall travel. That's really what OpenPATH is designed to do.

So, it would have to be on their phone. If there's time, we could actually try and talk to the car share people about integrating it into their app, 'cause OpenPATH is open source so, our code is there. If they have time, they can actually integrate it. If not, you might have to ask people to install two apps. But yeah, let's follow-up off-line. I'd love to be able to experiment with things that are other than eBikes, too.

Lauren Reichelt: Yeah. That's great. Yeah. Miriam has her hand up.

Miriam: Hi. I was wondering if this tool could be potentially used with – for example, fleets. That's one question I was wondering about. We have some fleets where they, for whatever reason – often times, financial – they don't have telematic software and so, I'm wondering – my first question is, can this be an option for those fleets with those lower budgets? And then, kind of to also piggy back off of that, we're working on EV charging station network planning, and again, trying to reach more underserved communities that might not have the capacity to get the open source from GitHub and do all of the customization work.

Can you tell me – are there some really easy ways for these people – different communities – to participate without doing all this customization?

Shankari: Yeah. So, that was the – so, first, for fleets, we have not used it for fleets before, but I don't see any reason why you couldn't use it for fleet. I mean, you'd basically buy like a $40.00 phone from the internet, install OpenPATH on it, and then, just leave the phone in your vehicle or whatever fleet thing you're monitoring, and then, it would give you a comprehensive picture of that vehicle's travel instead of a person's travel. It will all be the same mode so, it'll actually be much easier. OpenPATH is solving a harder problem, but it can certainly solve the easier problem of just having – sit in a vehicle. You don't even have to actually care about power drain at that point, because you can potentially even leave it plugged in, but that's a separate thing.

So, for the easy to use approach, this is sort of what we have in mind. So, we're working on getting all the pieces together at this point. I've talked to – to have this actually work at NREL, you need to get permissions from multiple different departments. There's cyber security. You need to get – if it's gonna run on a fed ramp certified medium cluster so, you know, we need to get permission from cyber.

We need to get permission from IT. We need to get permission from contracting. We get permission from legal. So, I'm lining all of those up, but hopefully, around the end of summer or so, I think we should have all our sort of ducks in a row and a process that we can have on a website. But basically, they would reach out to – if there's somebody who can sign a contract – like, a city or somebody like that – it would be free if they didn't want any customization.

If they wanted any customization, then it would be time for that customization. So, basically, what we have now for CanBikeCo – the energy e-bike one – we might make some minor changes to it. We'll make that available for free. You just have to sign an MOU with us. You'll get your own page – that will be like, city name.

You'll have a description of why you're collecting the data. People will just install OpenPATH in the stores, they'll scan a QR code, and then, they'll join that particular city's study. That's it.

Mirian: Okay. And – I think this is my last question. I'm sorry. When we are – Clean Cities coalition is working with a lot of small communities within our region – communities, say, between 1,000 to 10,000 people in the population – would you recommend that the Clean Cities coalitions – that we go ahead and set up our own standard MOU for our region or would you recommend that the individual small community set up their own MOU with NREL?

Shankari: I think it may be easier for you to do a group and then come to NREL – especially if it's really small communities, 'cause there's contracting overhead at NREL. But you know, I talked to contracting and they said we can do really small contracts, too. So, I think it maybe depends – it would be easier, but if they really want to have their own thing, then we can accommodate that as well.

Miriam: Thank you.

Shankari: And we're really interested in rural mobility. So, as you know, they're like, small communities, right? Yeah. And so, especially when we talk about the TSDC and the archival and so on. There is such little data about travel patterns in rural areas. There's actually a whole group – there's somebody on NREL who's focusing on rural mobility and it's like, there's not a lot of data around what people are actually doing there. And so, I think we'd be very interested in having that data for research.

Miriam: Okay. Thank you.

Shankari: Are there any other questions in the chat?

Lauren Reichelt: I don't see any other questions in the chat right now. Does anyone have any other questions for Shankari? Otherwise, I think we have some questions that we'd love to pose back to the group and have folks chime in. Okay. We will go that route until someone jumps in with another question.

Shankari, maybe you can pose one of these questions and we can try to get some responses as opposed to trying to tackle all five at once.

Shankari: Yeah. So, I guess in terms of – in the spirit of having a sort of open discussion and how this might be useful, we've already heard a couple of options – one about maybe a car share program or maybe understanding EV siting. Are there other ideas – even if you don't have a particular project in mind – any sort of brainstorming about urban planning decisions or transportation planning decisions like the examples that I gave that you'd like to share? I mean, maybe something that you already finished where you were like, "Wow. It would have been really good to have that kind of data."

Maggie: I wonder – again, this is Maggie, with Michigan Clean Cities, Greater Lansing area Clean Cities – one of the things that comes to mind is ahead of potential decisions to expand conventional transportation infrastructure – like, highways and roads, widening projects, or even new road building – I wonder if this kind of data would be helpful to show other ways, other modes, that can effectively meet the needs of residents and other users as opposed to the old way, which is just put a road in the middle of a neighborhood.

Shankari: Yeah. I'd be – I think – so, I think that moving from sort of program support to infrastructure support would be another sort of step forward for the platform as a whole, 'cause right now, we're like, "Well, we did e-bike programs. Now, maybe we'll do other kinds of programs." But then, after that, I think the next step might be to actually try and influence infrastructure decisions because one of the big reasons that I focused on this area for my work is we have so much data on cars, right? We have the infrastructure sensors, freeway sensors and so on that tell us how cars are going, how fast they're going, where they're going, and so on and so forth, but there is such little data on anything else that my concern was that what gets measured is what gets improved.

And so, if we could actually measure holistically, we would make better decisions going forward. And the built environment – I mean, there's a lot we can do with behavior change around incentives, but I think that, at some point, really, the behavior change has to come through land use decisions, right? Or the sort of infrastructure level decisions. 'Cause if something is easy and convenient and safe, then people will do it.

Lauren Reichelt: I like that. What gets measured gets improved.

Shankari: Yep.

Lauren Reichelt: Any other ideas for possible applications of this tool – whether it's a program locally, whether it's your program or a partner's program? Any other last thoughts as we are coming up on the hour here?

Shankari: Oh, go ahead. I was just gonna say – maybe one other question that you could think about in the sort of back of your mind before you reach out to us is what are the customizations that we should support? So, we cannot support everything, right? That's – e-bike is not gonna support everything and neither are we. But what are the high return on investment customizations that we should support at NREL? Maybe that people can access with a budget of 5K or something like that.

We're still envisioning that there would be a free option where you can't customize anything, but if people did want to customize, we're thinking somewhere on the order of 5K for some simple customizations. What should those customization be? So, I think that might be something else that will help us decide what we should prioritize in terms of getting this ready.

Lauren Reichelt: This is great. If anyone would like to reach out to Shankari – do you want to share your e-mail in the chat?

Shankari: Yeah. I should have actually put it in the last slide, huh? So –

Lauren Reichelt: And then, we do have one quick question that I think you can answer really swiftly. It was a follow-up from Miriam regarding rural communities. So, is the phone required to stay connected to high-speed data or is GPS sufficient if the internet or data connection drops?

Shankari: GPS is sufficient. So, I explicitly designed the app so that it does not have to be connected to the internet at all times. I was actually not thinking of rural communities at the time. I was thinking about power drain because actually having the radio on all the time to transmit and receive data is extremely energy intensive. The radio, if I remember right, is like, the second most intensive sensor component of the phone, just behind the screen.

So, it will be equivalent to having a screen on all the time. So, I just nixed that idea like, fairly early in the game. So, the phone – the app will cache data on the phone for as long as you have space on your phone. It will upload when it's connected to Wi-Fi. It would be good to have a SIM card, even if there is no data plan, because sometimes, it will triangulate using the cell towers to determine whether you've moved far enough to restart the trip.

So, when I've tested – I've actually tested with test phones that have no data plan. It's just a phone I bought off of Ebay with a cracked screen for $40.00 or something like that and it works. It works better if I put a SIM card in without a data plan, though.

Lauren Reichelt: All right. Great. So, I have a few people on here who are interested in connecting so, I can share that list with you afterwards, Shankari and we – and then, everyone, please, feel free to connect with Shankari via e-mail as well if you have ideas or suggestions for future iterations of this or customization. So, thank you all for joining the first EEMS Coffee with a Researcher webinar and we will share future webinars with you all when they have them scheduled. Have a great day, everyone.

Shankari: And if you want more details about how the app works and so on, just ping me. I will send you way more information than you want probably.

Lauren Reichelt: Thank you so much, Shankari.

Miriam: Great tool.