• 00:06:55 Bern Grush: Bern Grush, Urban Robotics Foundation (ISO Standards for robotics in public spaces)

• 00:08:56 Michael Schnuerle (OMF): Agenda and links for today https://github.com/openmobilityfoundation/mobility-data-specification/wiki/Web-conference-notes,-2026.06.04-(MDS-Working-Group)

• 00:09:31 Andrew Glass Hastings (OMF): Hi all! Andrew, the ED of the OMF. Welcome!

• If you haven’t already please change your name in Zoom to add your city, org, company name. We ask this for community building - so the OMF community on the call can get more familiar with the people and organizations that engage.

• 00:10:29 Michael Schnuerle (OMF): Calendar https://calendar.google.com/calendar/u/0/embed?src=openmobilityfoundation.org_g6gsaccjvijnmlhigfpj01ngp0@group.calendar.google.com&ctz=America/Los_Angeles

• 00:11:16 Grant Sivesind (Daro): Grant Sivesind from Daro along with @Grady Williams ! https://www.daro.city/

• 00:11:22 Andrew Glass Hastings (OMF): Replying to "Hi all! Andrew, the ..."

• If you are less familiar with Zoom click the 3 dots next to your name and click rename. Thanks all!

• 00:11:29 Michael Schnuerle (OMF): MDS 2.1 https://github.com/openmobilityfoundation/governance/wiki/Release-2.1.0

• 00:12:16 Aylene McCallum (OMF): Congrats Michael and all of the GitHub pull request creators and issue creators - your feedback and input into the process is invaluable!

• 00:12:51 David Von Stroh (Parametrix): Reacted to "Congrats Michael and..." with 👏

• 00:13:27 Andrew Glass Hastings (OMF): The OMF Board unanimously approved the MDS 2.1 release, which is a strong testament to all the work of the contributors! And a HUGE thanks to Micheal for his leadership moving the release to this point!

• 00:14:42 Michael Schnuerle (OMF): Crash data structure https://github.com/openmobilityfoundation/mobility-data-specification/blob/dev/data-types.md#incidents

• 00:20:33 Espen Johnsson: Can the zoom menu be removed?

• 00:22:32 Michael Schnuerle (OMF): Replying to "Can the zoom menu be..."

• I’ll ask at a break, but don’t want to interrupt right now. Thanks!

• 00:23:09 Espen Johnsson: Replying to "Hi all! Andrew, the ED of the OMF. Welcome!

• If you haven’t already please change your name in Zoom to add your city, org, company name. We ask this for community building - so the OMF community on the call can get more familiar with the people and organizations that engage."

• On then presentation, not on my side

• 00:30:56 Aylene McCallum (OMF): Great question John!

• 00:31:14 Andrew Glass Hastings (OMF): She is at Micromobility Europe

• 00:31:15 Oskar Jahr (Municipality of Bergen): Just wanted to comment that the analysis done to the demographics of the Cowboy user group and it's possible blindspots is really important, and something we all have to be mindful of. For much of the user group that MDS will encompass the demographic will be skewed one way or another compared to the regular population

• 00:31:34 Blaine Ebert: Reacted to "Just wanted to com..." with 👍

• 00:31:45 Maria Cortes: Reacted to "Just wanted to com..." with 👍

• 00:31:47 Aylene McCallum (OMF): Reacted to "Just wanted to comme..." with 👍

• 00:31:51 Raquel Corchado: Yes, I really appreciated that call out for equity concerns to have in mind

• 00:31:55 Raquel Corchado: Reacted to "Just wanted to com..." with 👍

• 00:32:25 Michael Schnuerle (OMF): Reacted to "Just wanted to comme..." with 👍

• 00:32:28 Oskar Jahr (Municipality of Bergen): Reacted to "Yes, I really apprec..." with 👍

• 00:32:36 Michael Schnuerle (OMF): Reacted to "Yes, I really apprec..." with 👍

• 00:32:37 Aylene McCallum (OMF): Replying to "Just wanted to comme..."

• The data is still pretty comprehensive and we are lucky enough that the bias in this data set is fairly transparent.

• 00:33:10 Bern Grush: Reacted to "Just wanted to comme..." with 👍

• 00:33:26 John Clary (City of Austin, TX): I have questions for you, Robbie, about your crash data sources but will ask them offline instead of going too far in the weeds. Thanks for your presentation!

• 00:33:27 Oskar Jahr (Municipality of Bergen): Replying to "Just wanted to comme..."

• Absolutely! It seems to be a treasure trove in how it will allow Amsterdam to fix danger hotspots in their bike network

• 00:34:01 Aylene McCallum (OMF): Reacted to "Absolutely! It seems..." with 👍🏻

• 00:34:19 Raquel Corchado: Robbir might just reach out to you to get connect with your collegues regarding operator engagement to share this data, thanks!

• 00:36:02 Robbie Vinogradov (City of Amsterdam): Reacted to "I have questions f..." with 👍

• 00:36:04 Robbie Vinogradov (City of Amsterdam): Reacted to "Robbir might just ..." with 👍

• 00:52:42 Aylene McCallum (OMF): Good point Michael... it is really important to remember and remind partners what this data is ultimately being used for (saving lives and preventing serious injury)

• 00:52:52 Andrew Glass Hastings (OMF): This is an excellent effort for next steps with MDS 2.1

• 00:54:46 Aylene McCallum (OMF): Of course!

• 00:55:32 Michael Schwartz, INRIX: I have to drop but thank you everyone — great discussion!

• 00:57:28 Bern Grush, Urban Robotics Foundation: Initial 4448-1: https://www.iso.org/standard/81068.html

• 00:58:02 Carl Hansen - Coco: Thank all!

• 00:58:09 Aylene McCallum (OMF): Reacted to "Thank all!" with 👍🏻

• 00:58:18 Andrew Glass Hastings (OMF): Replying to "Initial 4448-1: http..."

• Bern - feel free to share the upcoming URF webinar that you invited me to join!

• 00:58:20 Bern Grush, Urban Robotics Foundation: I am at bern@urbanroboticsfoundation.org for Journey Data Recorder work to date.

• 00:59:07 John Clary (City of Austin, TX): Nice to see ya’ll. Really good stuff—thanks!

• 00:59:10 Bern Grush, Urban Robotics Foundation: Replying to "Initial 4448-1: http..."

• https://www.urbanroboticsfoundation.org/event-details/robots-in-public-panel-discussion

• 00:59:20 Aylene McCallum (OMF): Reacted to "Nice to see ya’ll. R..." with ❤️

WEBVTT

1 00:00:03.670 --> 00:00:14.660 Michael Schnuerle (OMF): All right, welcome everybody to the OMF's MDS Working Group public meeting. Today, we're going to be talking about MDS 2.1 features, specifically

2 00:00:14.660 --> 00:00:26.550 Michael Schnuerle (OMF): crash data features that are in practice in a certain way, either to potentially be in MDS or helped inspire the work for MDS 2.1.

3 00:00:26.760 --> 00:00:31.709 Michael Schnuerle (OMF): My name is Michael Schnurle, I'm the Director of Open Source Operations for the OMF.

4 00:00:32.020 --> 00:00:46.839 Michael Schnuerle (OMF): Thanks for coming today. Go ahead and rename yourself in the participant list, and set your organization after your name, so we know who you're representing today. That would be really helpful for us and for note-taking.

5 00:00:47.920 --> 00:01:01.449 Michael Schnuerle (OMF): Mute yourself for now, but feel free to introduce yourself in the chat, whether it's your first time or you've been here a number of times. Feel free to say who you are, who you're with, and what you're excited about while you're here.

6 00:01:01.700 --> 00:01:20.329 Michael Schnuerle (OMF): Feel free to share links as well to relevant topics or projects. Use the chat to leave questions as we go, or use the raise hand feature. We'll have Q&A at different points throughout it as well. And this meeting is being recorded and will be published to the mailing list with notes.

7 00:01:22.230 --> 00:01:33.050 Michael Schnuerle (OMF): So, I'm going to give a little bit of an intro and do some announcements here, specifically about MDS 2.1 and the steering committee, and then we'll hear from Amsterdam and DDOT.

8 00:01:34.610 --> 00:01:46.780 Michael Schnuerle (OMF): So, to start off, the OMF is a global nonprofit created in 2019, city-led, city-founded, and, working with the private sectors to get to yes.

9 00:01:47.680 --> 00:02:02.149 Michael Schnuerle (OMF): We build everything in the open, lots of participation for both of our standards in public forums and online in public places, like GitHub and other forums, and we want to build useful things

10 00:02:02.390 --> 00:02:09.129 Michael Schnuerle (OMF): quickly, Nimbly for everyone to use, make the public space better.

11 00:02:09.300 --> 00:02:21.740 Michael Schnuerle (OMF): MDS is a data standard that enables right-of-way regulation, digital policy, geofencing, two-way communications. You can see our steering committee members on the right there.

12 00:02:22.050 --> 00:02:26.040 Michael Schnuerle (OMF): And our chairs from LA and Blue Systems in bold.

13 00:02:26.160 --> 00:02:38.809 Michael Schnuerle (OMF): And I'll talk a little bit more about the 2.1 release, which they had, and you had a hand in developing. So, MDS 2.1 now supports all the different modes and vehicles and services you see here.

14 00:02:39.310 --> 00:02:51.289 Michael Schnuerle (OMF): And, more cities are using these all the time. In fact, in DC, I just… earlier this week, I did testimony in front of Council about

15 00:02:51.490 --> 00:02:57.139 Michael Schnuerle (OMF): different taxis and TNCs and even autonomous vehicle.

16 00:02:57.260 --> 00:03:07.350 Michael Schnuerle (OMF): robotaxis, about how they can make sure that all of those modes operate on a level playing field, and MDS was mentioned a number of times.

17 00:03:08.900 --> 00:03:20.139 Michael Schnuerle (OMF): MDS is global. Over 200 commercial mobility service providers send MDS to cities or receive it from cities, and these companies operate in over 1,200 cities around the world.

18 00:03:21.650 --> 00:03:32.080 Michael Schnuerle (OMF): And our other data standards, CVS, we have different working groups for that one, and meetings for that one, but they work together to manage the entire public realm.

19 00:03:32.080 --> 00:03:46.549 Michael Schnuerle (OMF): They both do digital policy, data sharing, and metrics, can be used independently or cooperatively, and MDS is about vehicle location status incidents, which we'll talk about today, and trips. And CDS is about the physical infrastructure

20 00:03:46.690 --> 00:03:50.409 Michael Schnuerle (OMF): And, activity tracking at that infrastructure.

21 00:03:52.040 --> 00:03:57.739 Michael Schnuerle (OMF): And I will put the agenda in the chat today with a few links in there.

22 00:03:58.300 --> 00:04:02.839 Michael Schnuerle (OMF): And that's where the slides are and the recording will be.

23 00:04:03.910 --> 00:04:08.920 Michael Schnuerle (OMF): And I'd like to have Aileen just share a little bit about some of our members.

24 00:04:11.190 --> 00:04:25.840 Aylene McCallum (OMF): So, we can't do the work that we do, like hosting today's meeting, without the support of our members, so I just wanted to acknowledge and thank our Premier members, Blue Systems, Curb IQ, InRex, Passport, and Nemojo, thank you so much for your support.

25 00:04:25.840 --> 00:04:36.060 Aylene McCallum (OMF): And then, also our associate members. I will just point out that we've had a number of new members join in the last couple of months.

26 00:04:36.060 --> 00:04:47.519 Aylene McCallum (OMF): They're all listed here, but the most recent one, is Laz that I wanted to point out, and that's actually their work logo. I'll be adding their updated logo, they're just known as Laz now.

27 00:04:47.520 --> 00:04:56.840 Aylene McCallum (OMF): So thank you so much to all of our Premier and Associate members, for all the work you do to support, our work.

28 00:04:58.340 --> 00:05:03.749 Michael Schnuerle (OMF): Great, thank you, Eileen, thanks to our members, and of course, we have many, over 80

29 00:05:03.870 --> 00:05:05.790 Michael Schnuerle (OMF): Public sector members as well.

30 00:05:07.050 --> 00:05:22.650 Michael Schnuerle (OMF): We do have an OMF calendar where we track every conference, speaking engagement, public meeting, webinar, and so if you're… if you want to subscribe to that, that's a great way to know what we're doing and keep up to date.

31 00:05:22.770 --> 00:05:24.490 Michael Schnuerle (OMF): I'll put that in the chat.

32 00:05:28.460 --> 00:05:41.100 Michael Schnuerle (OMF): And then I want to talk a little bit about MBS 2.1. So, this is our highlight of our OMF deliverable approval process. It goes through this group here, the working groups and the steering committee.

33 00:05:41.360 --> 00:05:52.810 Michael Schnuerle (OMF): And a release candidate is made, recommended to the Tech Council. The Tech Council has already, with 2.1, released, reviewed and, approved and recommended that release.

34 00:05:52.810 --> 00:06:02.390 Michael Schnuerle (OMF): to the board, and the board has also had a chance to review and give feedback and consider the 2.1 release.

35 00:06:02.480 --> 00:06:06.610 Michael Schnuerle (OMF): And I'm happy to announce it's now approved. The board approved it.

36 00:06:06.780 --> 00:06:22.770 Michael Schnuerle (OMF): I think officially this week, and so you can read all about it. I'm not going to go into the features this time, but we have a great release plan. Maybe I'll put that in the chat, which has release notes and links to other things.

37 00:06:23.420 --> 00:06:28.189 Michael Schnuerle (OMF): There we go… And,

38 00:06:28.750 --> 00:06:40.319 Michael Schnuerle (OMF): Yeah, so what the next step is that I will be working to make sure everything is packaged up the right way, and then make it the official release on the MDS GitHub homepage.

39 00:06:41.470 --> 00:06:47.729 Michael Schnuerle (OMF): And I'd like to acknowledge, we'll have more acknowledgements when the release is finalized, but just a quick list of…

40 00:06:47.780 --> 00:07:05.349 Michael Schnuerle (OMF): people who created pull requests, so those are actual changes to the spec in GitHub, and there's a nice list of members and non-members there, and then GitHub issue creators as well, who raised an issue and then had something develop and be put into the release based on that, so thank you all.

41 00:07:06.760 --> 00:07:23.130 Michael Schnuerle (OMF): And, related to this, our steering committee is up for their annual nominations, so I just wanted to share this. We'll be sharing more information with our OMF members, but if you are an OMF member, you can choose to serve on the MDS Working Group Steering Committee.

42 00:07:23.240 --> 00:07:36.359 Michael Schnuerle (OMF): There will be a form mailed out soon, and you can nominate yourself, or talk to a coworker and nominate them. You can see our current members on the right. I hope that all of them

43 00:07:36.360 --> 00:07:51.200 Michael Schnuerle (OMF): nominate themselves again, because they've been fantastic. But there's no cap on the number of members of this group, so feel free to nominate and help weigh in on issues, run these, set the agendas for the meetings, like this one.

44 00:07:51.230 --> 00:07:53.389 Michael Schnuerle (OMF): And approved releases.

45 00:07:53.560 --> 00:07:56.390 Michael Schnuerle (OMF): So, thanks to our steering committee.

46 00:07:58.860 --> 00:08:08.370 Michael Schnuerle (OMF): All right, with that, I think that's all the intro and announcements. So we're going to get into two examples of MDS in practice around crash data.

47 00:08:08.590 --> 00:08:15.610 Michael Schnuerle (OMF): I'm gonna… I just have one slide to sort of level set a little bit. We've shared this information before.

48 00:08:15.700 --> 00:08:26.730 Michael Schnuerle (OMF): But basically, in MBS 2.1, there is a new endpoint, called Incidents, and Incidents can capture all the things you see here.

49 00:08:26.740 --> 00:08:47.379 Michael Schnuerle (OMF): where possible, where it's required, where it's regulated, where it's written into contracts and agreements. So it can capture the kinds of incidents you see here, unplanned stops for autonomous vehicles, maybe remote takeovers, ADS disengagements or engagements for other vehicles as well, like harsh braking and acceleration.

50 00:08:47.380 --> 00:08:54.970 Michael Schnuerle (OMF): Tipovers, maybe that's connected more to scooters, vandalism or theft incidents, near misses, which we'll hear about today a little.

51 00:08:54.970 --> 00:08:57.660 Michael Schnuerle (OMF): And collisions or crashes.

52 00:08:57.780 --> 00:09:16.069 Michael Schnuerle (OMF): And each of these gets basic information in the short term as available, so the location, date, time, severity, was there a medical response, accelerometer data, etc. And a link to, when available, a potential external report identifier.

53 00:09:16.510 --> 00:09:30.130 Michael Schnuerle (OMF): And then all of this is connected to MDS telemetry points, so you can see if they were on a trip or not, and this does work for all modes, when available for those vehicles. And I will also put a link…

54 00:09:30.780 --> 00:09:34.040 Michael Schnuerle (OMF): To, just the data structure here.

55 00:09:35.720 --> 00:09:36.930 Michael Schnuerle (OMF): In the chat.

56 00:09:37.390 --> 00:09:43.199 Michael Schnuerle (OMF): So you can see what it is, and then you'll have to click around the rest of the,

57 00:09:43.350 --> 00:09:46.019 Michael Schnuerle (OMF): Spec to see the endpoints and things like that.

58 00:09:48.150 --> 00:09:56.409 Michael Schnuerle (OMF): Alright, so, we have two presentations today. I'll… let me just tee them up a little. Amsterdam.

59 00:09:56.440 --> 00:10:11.840 Michael Schnuerle (OMF): has been working with crash and near-miss data for e-bikes for a while, and they are investigating, and we've talked to them about aligning their bespoke data standard into MDS 2.1, and we think it aligns really well. So they're going to be sharing

60 00:10:11.840 --> 00:10:18.650 Michael Schnuerle (OMF): about what Amsterdam is doing with that data, and how they get it, what they do with it, and why they need it.

61 00:10:18.720 --> 00:10:36.270 Michael Schnuerle (OMF): And then, DC's DDOT also, really was the inspiration for this addition to MDS 2.1, because they are asking for, crash and incident data for almost every mode that they can operate on, so…

62 00:10:36.370 --> 00:10:48.310 Michael Schnuerle (OMF): Either in practice now, or in the future, or potentially in the future, everything from scooters and bike share to personal delivery devices, which we'll hear about today.

63 00:10:48.310 --> 00:11:07.929 Michael Schnuerle (OMF): autonomous vehicles, car share, so all of it is hopefully going to be in the MBS 2.1 format now that they've brought the need here, and they have ordinances about this, and so we get to hear from them as well. So, we'll start first with Robbie from Amsterdam.

64 00:11:09.210 --> 00:11:10.949 Michael Schnuerle (OMF): Hi, I see you on camera.

65 00:11:11.660 --> 00:11:13.410 Robbie Vinogradov (City of Amsterdam): Yes, thank you, Michael.

66 00:11:13.700 --> 00:11:26.520 Michael Schnuerle (OMF): Yes, I'll stop sharing, and you can feel free to share your screen, and I'd say, if you want to do, maybe, thinking about time, 15 minutes or so, and we can have a little Q&A, and it'll be similar for DC.

67 00:11:28.550 --> 00:11:29.440 Robbie Vinogradov (City of Amsterdam): Sounds good.

68 00:11:31.270 --> 00:11:34.179 Robbie Vinogradov (City of Amsterdam): Swap the view, can you see my screen right now?

69 00:11:34.490 --> 00:11:35.700 Michael Schnuerle (OMF): Yeah, it looks good.

70 00:11:35.700 --> 00:11:36.910 Robbie Vinogradov (City of Amsterdam): Okay, great.

71 00:11:37.710 --> 00:11:55.430 Robbie Vinogradov (City of Amsterdam): So, hi everyone, I'm Roby Vinogradov, GIS analyst at Amsterdam's Innovation Department. Today, I'll walk you through how we have been working with bike crash and near crash data from Cowboy, a very popular e-bike brand in Amsterdam.

72 00:11:55.580 --> 00:12:02.770 Robbie Vinogradov (City of Amsterdam): And I'll share what we've learned so far, and try to connect it to where OMF is heading with the MDS Crash API.

73 00:12:04.810 --> 00:12:20.659 Robbie Vinogradov (City of Amsterdam): But first, let's start with a bit of context on why cycling safety matters so much in Amsterdam. It's probably no surprise that cycling is our dominant mode of transport. About 47 of all trips are made by bike in Amsterdam.

74 00:12:20.890 --> 00:12:25.480 Robbie Vinogradov (City of Amsterdam): But cyclists are also overrepresented in traffic injuries.

75 00:12:25.670 --> 00:12:31.040 Robbie Vinogradov (City of Amsterdam): Roughly half of our ambulance attendant traffic accidents involve a cyclist.

76 00:12:31.480 --> 00:12:38.949 Robbie Vinogradov (City of Amsterdam): And on top of that, e-bikes are becoming more and more common. In 2023, they already made up

77 00:12:39.300 --> 00:12:46.659 Robbie Vinogradov (City of Amsterdam): 20-30% of bikes on our roads, and more than half of all new bikes sold are e-bikes.

78 00:12:47.890 --> 00:12:55.779 Robbie Vinogradov (City of Amsterdam): They go faster, and therefore residents are increasingly worried about safety, especially because of these e-bikes.

79 00:12:57.620 --> 00:13:12.219 Robbie Vinogradov (City of Amsterdam): And the way we currently handle cycling safety is mainly through black spots. And a black spot is defined as a road section where at least 3 injury accidents have been reported within 3 years.

80 00:13:12.260 --> 00:13:17.869 Robbie Vinogradov (City of Amsterdam): Usually by the police. And once a black spot is identified.

81 00:13:18.000 --> 00:13:23.310 Robbie Vinogradov (City of Amsterdam): A traffic safety team visits the location to assess and suggest improvements.

82 00:13:23.890 --> 00:13:30.849 Robbie Vinogradov (City of Amsterdam): It works, but it's slow and very reactive. We only act after multiple people have already been hurt.

83 00:13:32.710 --> 00:13:44.740 Robbie Vinogradov (City of Amsterdam): However, serious accidents don't just happen out of nowhere. They are usually preceded by many smaller incidents and near misses, and this principle is called the Heinrich Safety Pyramid.

84 00:13:44.880 --> 00:13:53.130 Robbie Vinogradov (City of Amsterdam): At the top of the pyramid, you'll find the more fatal and serious accidents, and this is where the police data lives, which we currently have.

85 00:13:53.270 --> 00:13:56.839 Robbie Vinogradov (City of Amsterdam): So, the reported crashes, think of ambulance calls.

86 00:13:57.210 --> 00:14:04.019 Robbie Vinogradov (City of Amsterdam): It's important data, but it only captures instances that are serious enough to be reported.

87 00:14:04.330 --> 00:14:18.490 Robbie Vinogradov (City of Amsterdam): Whereas, lower in the pyramid, you'll find minor incidents and near misses and other unsafe situations. And this is where the cowboy data sits. It picks up questions and files that don't reach the police.

88 00:14:18.690 --> 00:14:23.220 Robbie Vinogradov (City of Amsterdam): And it has the potential to provide a much bigger signal to work with.

89 00:14:25.960 --> 00:14:45.930 Robbie Vinogradov (City of Amsterdam): And we can already see that the signal is growing pretty fast. The charts below, the chart below shows the bike incidents reported by Cowboy in orange, and the police in blue over time. And in just a few years, Cowboy reported more bike incidents than the police in 2024.

90 00:14:45.930 --> 00:14:50.139 Robbie Vinogradov (City of Amsterdam): And that's just from one provider, which suggests that it's a much…

91 00:14:50.310 --> 00:14:54.690 Robbie Vinogradov (City of Amsterdam): A larger number of incidents that traditional police data does not capture.

92 00:14:55.280 --> 00:15:06.349 Robbie Vinogradov (City of Amsterdam): So, with the crash data from providers such as Cowboy, we can move down the pyramid and act earlier before the serious accidents happen.

93 00:15:09.130 --> 00:15:21.640 Robbie Vinogradov (City of Amsterdam): So, as I mentioned earlier, Cowboy is a popular e-bike brand in Amsterdam. Cowboy riders made up around 7 million trips in a city between 2021 and 2024.

94 00:15:21.640 --> 00:15:29.569 Robbie Vinogradov (City of Amsterdam): And their bikes have a built-in crash detection system. There are sensors in the wheel, frame, and bottom brackets.

95 00:15:29.640 --> 00:15:32.480 Robbie Vinogradov (City of Amsterdam): That, together, detect when a rider falls.

96 00:15:32.680 --> 00:15:42.130 Robbie Vinogradov (City of Amsterdam): And if the rider doesn't dismiss the alert that will appear on their phone after a crash, within 60 seconds, their emergency contacts are notified automatically.

97 00:15:42.760 --> 00:15:45.310 Robbie Vinogradov (City of Amsterdam): And about a year ago.

98 00:15:45.680 --> 00:15:56.400 Robbie Vinogradov (City of Amsterdam): they shared their data with us as a one-time export in the form of just an Excel spreadsheet, so we could see what this data could offer for a city like Amsterdam.

99 00:15:56.850 --> 00:16:02.789 Robbie Vinogradov (City of Amsterdam): And the data came in two parts. So, first, we received the crash data.

100 00:16:03.040 --> 00:16:07.579 Robbie Vinogradov (City of Amsterdam): So, for every detected incident, we got the timestamp.

101 00:16:07.900 --> 00:16:21.260 Robbie Vinogradov (City of Amsterdam): the GPS location, the vehicle speed 3 seconds before, during, and 3 seconds after the crash, and a severity score, minor or major, based on the feedback of the rider.

102 00:16:22.370 --> 00:16:35.220 Robbie Vinogradov (City of Amsterdam): Second, we also received trip data, which are the hourly counts of rides started, which we use to normalize the crash rate and see where incidents are genuinely overrepresented.

103 00:16:38.360 --> 00:16:45.630 Robbie Vinogradov (City of Amsterdam): And since then, we've built a bike crash dashboard that combines the cowboy and the police data in one place.

104 00:16:46.220 --> 00:16:56.959 Robbie Vinogradov (City of Amsterdam): As you can see, it has a lot of filters and also views, so I'll just highlight a couple of things. And sorry, I didn't take the time to translate everything into English for this presentation, but I'll walk you through it.

105 00:16:57.360 --> 00:17:01.440 Robbie Vinogradov (City of Amsterdam): So this view shows the cowboy crash data by speed.

106 00:17:02.080 --> 00:17:07.530 Robbie Vinogradov (City of Amsterdam): The bottom left chart shows instant counts by hour of the day in orange.

107 00:17:07.890 --> 00:17:12.340 Robbie Vinogradov (City of Amsterdam): With the red line showing the average speed before the incident.

108 00:17:12.480 --> 00:17:24.000 Robbie Vinogradov (City of Amsterdam): And as you can see, the pre-incident speed is, for example, highest during the morning rush hour, around 20 kilometers an hour, which also makes sense with commuters.

109 00:17:24.240 --> 00:17:25.740 Robbie Vinogradov (City of Amsterdam): Riding fast.

110 00:17:26.560 --> 00:17:38.210 Robbie Vinogradov (City of Amsterdam): The bottom right chart shows that incident counts rise as speed increases, with a peak around, well, let's say, 25 kilometers an hour before dropping down.

111 00:17:38.430 --> 00:17:45.989 Robbie Vinogradov (City of Amsterdam): And that correlation makes sense, but higher speeds also go together with busier times of the day and more traffic.

112 00:17:46.330 --> 00:17:58.330 Robbie Vinogradov (City of Amsterdam): So, we cannot simply conclude that speed causes crashes based on the chart, but, I mean, we can say that speed data adds useful context that police records almost never capture.

113 00:18:01.960 --> 00:18:15.469 Robbie Vinogradov (City of Amsterdam): Another thing that makes this data valuable is how it also reflects real-world conditions. This chart, for example, shows cowboy crash detections day by day, it's the red line.

114 00:18:15.620 --> 00:18:21.250 Robbie Vinogradov (City of Amsterdam): And we have overlaid it with the minimum daily temperature, the blue line, in degrees Celsius.

115 00:18:21.420 --> 00:18:39.540 Robbie Vinogradov (City of Amsterdam): And the vertical yellow and the two orange lines mark a, what we call in the Netherlands, a code yellow and a code orange, which are basically weather alerts for, in this case, black ice. And as we can see, each time there was a warning, the temperature drops below the freezing point, and the crash rates

116 00:18:39.550 --> 00:18:45.129 Robbie Vinogradov (City of Amsterdam): Spiked sharply, sometimes from a few per day to well over 30.

117 00:18:45.510 --> 00:18:58.670 Robbie Vinogradov (City of Amsterdam): And also, this is a kind of insight that's very practical, because knowing exactly where most incidents cluster during icy conditions in a city could help the city to decide where to prioritize

118 00:18:58.820 --> 00:19:00.399 Robbie Vinogradov (City of Amsterdam): Gritting, for example.

119 00:19:05.530 --> 00:19:14.719 Robbie Vinogradov (City of Amsterdam): If you only look at the raw crash numbers, busy areas will usually always pop up, simply because more people cycle there.

120 00:19:14.840 --> 00:19:20.350 Robbie Vinogradov (City of Amsterdam): For example, a major junction in the city center might have

121 00:19:20.500 --> 00:19:24.099 Robbie Vinogradov (City of Amsterdam): Let's say, 10 times more crashes than a quiet residential street.

122 00:19:24.200 --> 00:19:27.770 Robbie Vinogradov (City of Amsterdam): But if it also has 10 times more cyclists.

123 00:19:28.030 --> 00:19:31.390 Robbie Vinogradov (City of Amsterdam): The actual risk per ride is basically the same.

124 00:19:31.780 --> 00:19:39.989 Robbie Vinogradov (City of Amsterdam): So, raw crest numbers mainly show where cycling happens. They do not always reflect where cycling is the most dangerous.

125 00:19:40.340 --> 00:19:46.830 Robbie Vinogradov (City of Amsterdam): And the map in the top left shows the raw cowboy crash locations per hexagon cell.

126 00:19:47.840 --> 00:20:03.340 Robbie Vinogradov (City of Amsterdam): And the map in the top right shows cycling intensity based on Strava e-bike data, also for Exconcel, and this shows where people cycle the most, and we also check this data with our own bike counting sensors to see if these intensities align.

127 00:20:03.780 --> 00:20:18.919 Robbie Vinogradov (City of Amsterdam): And when you divide the number of crashes by the number of rides per hexagon cell, you end up with a map at the bottom. So, what you see over there are the number of crashes per… a thousand rides per hexagon cell.

128 00:20:19.070 --> 00:20:23.779 Robbie Vinogradov (City of Amsterdam): And we also used a minimum threshold, or a ride threshold.

129 00:20:23.940 --> 00:20:27.699 Robbie Vinogradov (City of Amsterdam): So we would only look at places with enough cycling volume.

130 00:20:27.980 --> 00:20:34.669 Robbie Vinogradov (City of Amsterdam): And this map already gives us a much better view of the true risk hotspots in the city. Some areas

131 00:20:34.750 --> 00:20:49.540 Robbie Vinogradov (City of Amsterdam): that maybe barely showed up in the raw data in the top left. Some of the look high risk, whereas other areas that looked dangerous at first turn out to be less risky once we take cycling density into account.

132 00:20:50.970 --> 00:20:59.339 Robbie Vinogradov (City of Amsterdam): And I think this is also why a crash API that also supports trip data, as MDS does, is very valuable for cities such as Amsterdam.

133 00:21:04.290 --> 00:21:22.769 Robbie Vinogradov (City of Amsterdam): There is, however, a limitation to keep in mind when working with this type of data. For example, when you look at the incidence by area, you'll notice that the cowboy data concentrates in the dense inner city parts, like the areas Centrum and Jad.

134 00:21:23.050 --> 00:21:39.430 Robbie Vinogradov (City of Amsterdam): While the outer neighborhoods, further from the center, areas like Ost, New West, and North, are often less wealthy and show up a lot less in cowboy's data, even though the police numbers, here in blue, are almost just as high.

135 00:21:39.590 --> 00:21:46.739 Robbie Vinogradov (City of Amsterdam): And that's a geographic bias that reflects where cowboy users are, and not necessarily where cycling is most dangerous.

136 00:21:48.370 --> 00:22:02.569 Robbie Vinogradov (City of Amsterdam): The chart below, showing the number of incidents per hour, tells a similar story. Compared to the police data, cowboy incidents are overrepresented at night.

137 00:22:02.870 --> 00:22:07.889 Robbie Vinogradov (City of Amsterdam): Which could suggest that Cabo users are more likely to write during those hours.

138 00:22:08.160 --> 00:22:17.719 Robbie Vinogradov (City of Amsterdam): But when we look at the trip data separately, we see that the number of trips at night is actually quite low, which means that the incident rate is disproportionately high.

139 00:22:17.920 --> 00:22:19.510 Robbie Vinogradov (City of Amsterdam): So you can imagine that

140 00:22:19.700 --> 00:22:28.629 Robbie Vinogradov (City of Amsterdam): Alcohol use, the darker conditions, and or fatigue are all possible contributing factors to this, observation.

141 00:22:29.630 --> 00:22:39.620 Robbie Vinogradov (City of Amsterdam): And together, these patterns show that cowboy data does have some blind spots. It over-represents younger cyclists and trips in the inner city.

142 00:22:40.200 --> 00:22:55.719 Robbie Vinogradov (City of Amsterdam): So, even though the cowboy data is still very valuable, it shows us that it only gives us a part of the picture. And besides that, we're missing data like breaking behavior for now, which, for example, could help better review near-misses.

143 00:22:59.020 --> 00:23:14.879 Robbie Vinogradov (City of Amsterdam): So, what we really need, and this is the last slide of this presentation, is a standardized way to collect this kind of safety or crash data from different mobility providers, not just from one brand, and I think this is really where a crash API build on MDS could be very interesting.

144 00:23:14.970 --> 00:23:20.720 Robbie Vinogradov (City of Amsterdam): And that's also why MSEM is working with the OMF to explore how this crash API can be used in practice.

145 00:23:20.810 --> 00:23:27.049 Robbie Vinogradov (City of Amsterdam): And, yeah, I think so far we are really happy to see where MBS 2.1 is going.

146 00:23:30.170 --> 00:23:39.190 Robbie Vinogradov (City of Amsterdam): If you have any questions about Cubway data, or if you'd like a more in-depth demo of the crash dashboard, feel free to reach out via email. Thank you.

147 00:23:40.400 --> 00:23:41.549 Michael Schnuerle (OMF): Thank you, Robbie.

148 00:23:42.960 --> 00:23:49.429 Michael Schnuerle (OMF): I… I wanted… I don't know if you're going to keep sharing, but at the bottom of the screen, there's a little Zoom notification.

149 00:23:49.600 --> 00:23:51.870 Michael Schnuerle (OMF): I don't know if you could hide that.

150 00:23:52.160 --> 00:24:08.239 Michael Schnuerle (OMF): But, yeah, great presentation. I know we met in person and talked about this, and you… you have a lot more to share about this. I think you kept it pretty tight for this meeting, but if people are interested, yeah, please reach out to Roby in Amsterdam.

151 00:24:08.730 --> 00:24:17.159 Michael Schnuerle (OMF): We… we feel like, and we're gonna work more on this together, but we feel like MBS 2.1 does align with

152 00:24:17.280 --> 00:24:30.510 Michael Schnuerle (OMF): you know, at least 90%, maybe a lot more of what you're looking for, and so we want to make sure, and make sure that any gaps are addressed. But what you're calling the crash API,

153 00:24:30.510 --> 00:24:39.170 Michael Schnuerle (OMF): we're currently calling the Incidents API, because we decided during the release to make it a little bigger than crashes, like I showed in that slide.

154 00:24:39.260 --> 00:24:45.490 Michael Schnuerle (OMF): But we're really excited about it. Does anyone have questions for Roby or Amsterdam?

155 00:24:46.020 --> 00:24:47.969 Michael Schnuerle (OMF): About this great work.

156 00:24:52.830 --> 00:25:07.710 Raquel Corchado: I'll just say, from here from Chicago, this is really… this is really great, and it's… it's… it's very useful to see how it can be used even for, like, weather conditions, and, like, which areas to prioritize based on, like, near crashes. So, yeah, just… just great work. I really enjoyed the presentation, thanks.

157 00:25:08.900 --> 00:25:09.650 Robbie Vinogradov (City of Amsterdam): Thank you.

158 00:25:10.440 --> 00:25:19.470 John Clary (City of Austin, TX): Hey, this is John Clary at the City of Austin. Again, yeah, really appreciated this presentation. Really, really neat to see your process.

159 00:25:19.670 --> 00:25:24.099 John Clary (City of Austin, TX): I laughed when I saw Cowboy as the name of the e-bike renter, because,

160 00:25:24.200 --> 00:25:33.729 John Clary (City of Austin, TX): well, yeah, in Texas, I think. When I saw the name of your presentation, I thought we were just calling e-bike riders cowboys. So that was kind of funny, but,

161 00:25:34.140 --> 00:25:39.159 John Clary (City of Austin, TX): I'm… yeah, I guess I'm kind of curious just how your relationship with that vendor, or the…

162 00:25:39.350 --> 00:25:49.190 John Clary (City of Austin, TX): service who op… the folks who operate Cowboy, maybe that's a public entity, but how y'all started that partnership of the data… data sharing.

163 00:25:50.750 --> 00:26:04.499 Robbie Vinogradov (City of Amsterdam): Yeah, so I wasn't involved in that part. That was my colleague Gemma Schreppers, who is usually in this working group. I'm not sure if she is in this group right now, but she's the best person to answer that question.

164 00:26:07.440 --> 00:26:08.100 John Clary (City of Austin, TX): Thank you.

165 00:26:08.100 --> 00:26:14.289 Michael Schnuerle (OMF): Can I… can I… yeah, she's a… she's been on our board as well, and…

166 00:26:14.430 --> 00:26:17.599 Michael Schnuerle (OMF): She will be able to answer that. Can I ask, though.

167 00:26:17.950 --> 00:26:27.869 Michael Schnuerle (OMF): Cowboy, the company, though, is a… it's a privately… like, people… private residents buy the Cowboy bikes from the private company, Cowboy, correct?

168 00:26:28.100 --> 00:26:34.019 Michael Schnuerle (OMF): And so, it's not like you have a relationship as the city directly with Cowboy the company.

169 00:26:34.450 --> 00:26:35.170 Robbie Vinogradov (City of Amsterdam): Nope.

170 00:26:35.170 --> 00:26:52.699 Michael Schnuerle (OMF): Right. And so they're… they… did they come to you then? Maybe you don't know this, but did they come to you to share, this information? Say, like, look, this is the sort of anonymized information we're getting, or do you know a little bit more about that back… background?

171 00:26:53.350 --> 00:27:12.639 Robbie Vinogradov (City of Amsterdam): I think it was actually during a conference when Hema and Tangi, the… I think he's now the former CTO, CEO, I'm not sure, of Cowboy, were in a meeting when they were talking about, crashes, and then Tangi actually told Shema that they were sitting on this

172 00:27:12.910 --> 00:27:22.189 Robbie Vinogradov (City of Amsterdam): cool data that they could share with us, and I think that's how the whole cooperation was initiated.

173 00:27:23.030 --> 00:27:23.670 Michael Schnuerle (OMF): Right.

174 00:27:26.390 --> 00:27:35.640 Andrew Glass Hastings (OMF): Yeah, I think, Robbie, I think from my previous conversations with Hema, I think you're absolutely right, and it was… always struck me as a really opportune example of…

175 00:27:35.640 --> 00:27:54.449 Andrew Glass Hastings (OMF): You have a private company that recognized through a set of conversations that they have data that could be really interesting to compile with existing, kind of, crash-related data that the city has to, to build out a clearer idea of what's happening with crashes and incidents on the street.

176 00:27:54.450 --> 00:28:10.469 Andrew Glass Hastings (OMF): doing it voluntarily. Like, no one, no one forced Cowboy to do this. They saw, they saw, the advantage of, of sharing their data with the city, through this, through this mechanism. So, it was kind of, a somewhat unique and, and really interesting example in that case.

177 00:28:10.930 --> 00:28:11.490 Robbie Vinogradov (City of Amsterdam): Yep.

178 00:28:16.270 --> 00:28:18.180 Michael Schnuerle (OMF): Any final questions or thoughts?

179 00:28:24.720 --> 00:28:30.820 Michael Schnuerle (OMF): All right, if you think of some others as the meeting continues, feel free to leave it in the chat.

180 00:28:30.950 --> 00:28:47.759 Michael Schnuerle (OMF): And, also, the slides… we will add these slides to our slide deck, so they'll be shared. So, if anything was obscured by the Zoom icons, then you'll be able to see it in the final shared slide. So, thank you very much for that presentation.

181 00:28:49.140 --> 00:28:49.930 Robbie Vinogradov (City of Amsterdam): You're welcome.

182 00:28:51.720 --> 00:28:53.270 Michael Schnuerle (OMF): Alright, next step.

183 00:28:53.440 --> 00:29:05.749 Michael Schnuerle (OMF): I won't share my screen yet, but next up we have DCDOT. I think we have Stephanie, both Stephanie and, some other folks from DDOT here to present, maybe.

184 00:29:09.810 --> 00:29:11.740 Stephanie Dock (DDOT | DC): Could I drag other people in?

185 00:29:11.930 --> 00:29:16.039 Michael Schnuerle (OMF): Maybe. Oh, maybe not. Maybe it's just you. I thought I saw some DC people.

186 00:29:16.320 --> 00:29:22.240 Stephanie Dock (DDOT | DC): If there are others from DDOT on, feel free to pipe up for whatever.

187 00:29:22.600 --> 00:29:23.960 Stephanie Dock (DDOT | DC): I am…

188 00:29:24.260 --> 00:29:40.809 Stephanie Dock (DDOT | DC): I say wrong here. So, hi folks. Stephanie Doc, I manage the Innovation Branch at DDOT. We're responsible for the oversight, amongst other things, of personal delivery devices, and we have been preparing to reopen our permitting process for robots.

189 00:29:40.810 --> 00:29:49.730 Stephanie Dock (DDOT | DC): Our council, in its infinite wisdom, originally set a weight limit of 90 pounds for an unladen device, and that has very effectively limited

190 00:29:49.730 --> 00:29:54.240 Stephanie Dock (DDOT | DC): Who can operate here, because 90 pounds is pretty small.

191 00:29:54.330 --> 00:30:04.079 Stephanie Dock (DDOT | DC): So we've, KiwiBot has been here, working up on the Howard University campus. That has required a permit from us, because most of that campus does not own its own sidewalks, we own them.

192 00:30:04.170 --> 00:30:16.109 Stephanie Dock (DDOT | DC): But, Council has recent… last year, took action to temporarily raise the WIT limit, and I just got word they'll actually be doing the markup on the permanent bill in early July.

193 00:30:16.340 --> 00:30:26.520 Stephanie Dock (DDOT | DC): So it looks like the weight limit might stay put at a higher weight of, I believe it's 275 or higher if we raise it in, regulation.

194 00:30:26.600 --> 00:30:36.849 Stephanie Dock (DDOT | DC): But in the meantime, we have been rewriting our terms and conditions in preparation for an expansion of operators. We have interest from several.

195 00:30:36.950 --> 00:30:56.100 Stephanie Dock (DDOT | DC): our permit is back open to those who've been in touch with us, and we… if all goes well, may begin to see additional operators out on the streets in July. So we're a bit in a moment of experimentation on our terms and conditions, because our term… our permit cycle runs calendar… on a calendar year basis.

196 00:30:56.100 --> 00:31:11.239 Stephanie Dock (DDOT | DC): So we are going to issue permits, basically halfway through the year, and then have a kind of an opportunity to revisit what we've put into those terms in preparation for the issuance of the next set of permits, or the renewal of those permits in early 2027.

197 00:31:11.310 --> 00:31:23.730 Stephanie Dock (DDOT | DC): So, this is, MDS 2.1 is coming at a timely moment where, we're not yet going to be requiring, the reporting, but are very interested in seeing if, as this

198 00:31:24.350 --> 00:31:35.679 Stephanie Dock (DDOT | DC): cycle gets going, if it works for our operators, to be able to move over to that with an eye to either requiring it or providing it… definitely right now, providing it as an option.

199 00:31:36.590 --> 00:31:40.930 Stephanie Dock (DDOT | DC): I was just gonna pull up, this is…

200 00:31:41.520 --> 00:31:57.120 Stephanie Dock (DDOT | DC): not fancy, and I guess I should put it in the presentation of… because this was our walkthrough of the very dry terms and conditions draft that we had, that a few changes are being made, and this group doesn't need the full walkthrough on all the

201 00:31:57.320 --> 00:32:06.920 Stephanie Dock (DDOT | DC): all the ins and outs of everything that we were going to be doing, though if you would like to see the terms, we'll be happy to share them once they're, legally sufficient. They're in final review right now, but…

202 00:32:07.300 --> 00:32:22.330 Stephanie Dock (DDOT | DC): For what this particularly relates to, we do require data and reporting, much of which we are hoping that MDS can answer for us. So a couple that come through, we request that all… we require that all incidents be reported to DDOT within 24 hours.

203 00:32:22.330 --> 00:32:31.399 Stephanie Dock (DDOT | DC): And in that case, an incident for this context is an injury to a person or an animal, damage to public space, or any sort of

204 00:32:31.660 --> 00:32:36.179 Stephanie Dock (DDOT | DC): collision with traffic. So it's pretty broadly defined.

205 00:32:36.440 --> 00:32:55.090 Stephanie Dock (DDOT | DC): We basically just want to know what'd you run into, and when did you do it, and was anybody hurt? And so, a lot of this, as Michael mentioned, the revival of the incident for the bringing back to life the conversation around the incident feed was in part because of what we are requiring, both here on the permit side for sidewalk delivery robots, and then also

206 00:32:55.090 --> 00:33:07.619 Stephanie Dock (DDOT | DC): looking at our autonomous vehicle permitting, should we get our rulemaking finalized, ever, we would be requiring very similar information from autonomous vehicles. I also… this slide was appropriate because it was all of our data and reporting

207 00:33:07.710 --> 00:33:13.029 Stephanie Dock (DDOT | DC): I'll note that a lot of what we have here, we intend to receive as an MDS

208 00:33:13.350 --> 00:33:20.589 Stephanie Dock (DDOT | DC): data point, rather than an actual… at the moment, I think we're requiring both the report and the data point. There's a whole story behind that.

209 00:33:20.820 --> 00:33:27.530 Stephanie Dock (DDOT | DC): Part of it is us figuring it out, and there's some quirks in there about what's reported, thanks to Council and its definitions.

210 00:33:28.050 --> 00:33:39.340 Stephanie Dock (DDOT | DC): That it is easier to surface if the companies help tell us them. But the… so the number of devices, number of trips, all of that, not surprisingly, will come directly through MDS.

211 00:33:39.340 --> 00:33:48.580 Stephanie Dock (DDOT | DC): The other piece that I think does relate back to the incidents and the broader definition that's being used is we do require them to tell us when there are cases where the PDDs had to come to an off-roadway stop.

212 00:33:48.700 --> 00:34:05.939 Stephanie Dock (DDOT | DC): So, in what might be news to any of the operators that are on the line, if you would like to provide this information via MDS, and can do so to meet the full requirement, then you don't have to send me an email, necessarily. We can work it out. But since MDS 2.1 is not

213 00:34:05.940 --> 00:34:24.560 Stephanie Dock (DDOT | DC): quite yet the full standard, and we are still figuring out our systems on our side, and I know our intermediaries are also figuring out how to surface this information for us. We're kind of in this in-between state, where I think we're going to ask for information and try to figure out what we can get via MDS instead of having to have a direct exchange.

214 00:34:24.659 --> 00:34:29.750 Stephanie Dock (DDOT | DC): Or rather, make it the API exchange, not an email or a monthly report.

215 00:34:30.330 --> 00:34:34.260 Stephanie Dock (DDOT | DC): So, michael, I didn't have much more than that.

216 00:34:35.000 --> 00:34:35.960 Stephanie Dock (DDOT | DC): So…

217 00:34:35.969 --> 00:34:42.409 Michael Schnuerle (OMF): No, that's a great overview. I'm hoping people have some questions for you.

218 00:34:43.499 --> 00:34:58.529 Michael Schnuerle (OMF): Could you… maybe a question which I sort of mentioned is, and you mentioned a little bit, but could you go into more detail about how you're thinking about using MDS specifically for crash and incident information, and for which

219 00:34:58.799 --> 00:35:02.569 Michael Schnuerle (OMF): Sort of modes and services that are operating in the public realm.

220 00:35:03.150 --> 00:35:12.820 Stephanie Dock (DDOT | DC): Yeah, so the two areas that my program covers are autonomous vehicles and sidewalk delivery robots, or PD… personal delivery devices. So we'll be…

221 00:35:13.080 --> 00:35:19.550 Stephanie Dock (DDOT | DC): The terms and conditions for the PDDs do require MDS, or the updated version that

222 00:35:19.640 --> 00:35:32.800 Stephanie Dock (DDOT | DC): everyone will be moving on to come probably July 1, will require that. And then, if we get our regulatory structure in place for autonomous vehicles, we have written in a requirement for MDS.

223 00:35:32.890 --> 00:35:44.330 Stephanie Dock (DDOT | DC): As it stands right now, our AV companies that are testing with their ADS and their automated driving system engaged are required already to report crashes to us, and so for any of the…

224 00:35:44.520 --> 00:36:03.369 Stephanie Dock (DDOT | DC): AV companies on the line, if you would like to report your crashes via MDS, we are happy to receive it that way as well, and be an early test case of some of this. How does it work? I should have brought up the AV requirements, but they are already, even without the permits, required to do that, we just aren't mandating MDS at this point.

225 00:36:03.510 --> 00:36:21.550 Stephanie Dock (DDOT | DC): But they have a very similar set of requirements to what the PDDs do. Pretty much any crash qualifies. They have to let us know. In that case, crashes are within 12 hours on the AVs. Thank you, Council, for that timeline, and updates within 5 days, with any, like, beyond…

226 00:36:21.770 --> 00:36:27.439 Stephanie Dock (DDOT | DC): the very minimum of what has occurred. We have had a few crashes reported to us.

227 00:36:27.680 --> 00:36:40.560 Stephanie Dock (DDOT | DC): We have two testing entities here, and yeah, well, then others that have come through, but, only two that have engaged the ADS. And so, there is a real opportunity to kind of workshop

228 00:36:40.820 --> 00:36:55.600 Stephanie Dock (DDOT | DC): how well this is working before it becomes a permit mandate, and that's kind of our thought, is to sequence. TNCs and taxis do not currently come in via MDS. As you mentioned, you testified at the council roundtable on this.

229 00:36:55.600 --> 00:37:02.439 Stephanie Dock (DDOT | DC): We are definitely pushing that we think everyone should be held to the same standard, and that there's no reason that the TNCs shouldn't.

230 00:37:02.770 --> 00:37:10.129 Stephanie Dock (DDOT | DC): also have to report comparable data, but I don't know that there's been as much incident discussion on that side.

231 00:37:10.230 --> 00:37:12.829 Stephanie Dock (DDOT | DC): And it would change things up, and

232 00:37:12.840 --> 00:37:31.780 Stephanie Dock (DDOT | DC): I have not chatted with the… I don't know if anybody from the Shared Fleet Devices is here, but I don't know that we've talked about it for that or for Car Share specifically, but I think as we start to filter it in and learn how to work with it and feed it into our systems, if we get some takers on testing it, I think we can continue to think about how that gets built into other program areas.

233 00:37:33.550 --> 00:37:45.159 Michael Schnuerle (OMF): And, off… I know you've shared this before in prior meetings, but off the top of your head, what other systems are using MDS at the moment, whether it's for crash data or not? You've got, like, bike share, scooter share…

234 00:37:46.210 --> 00:37:57.200 Stephanie Dock (DDOT | DC): So, all of our jobless bike share, bike… I believe the scooter… Bike and Scooter Share are definitely doing it. Not our capital bike share system, but the e-bike portion of it also comes in that way.

235 00:37:57.320 --> 00:38:02.029 Stephanie Dock (DDOT | DC): I think. Yes, because I can see it in Ride Report. And then,

236 00:38:02.240 --> 00:38:19.399 Stephanie Dock (DDOT | DC): car share, I believe, comes in. That's not a huge feed, and then AVs will be expected more broadly to use it. And then we are working on talking about it with Council, but there's a bit of a push going on to suggest that

237 00:38:19.720 --> 00:38:37.439 Stephanie Dock (DDOT | DC): that we think that Council should revisit how taxis and TNCs report their data. The taxi companies here do already provide breadcrumb data, so their… their work on that predates MDS 2.0 by quite a bit. That goes back to probably before 2018.

238 00:38:37.550 --> 00:38:39.700 Stephanie Dock (DDOT | DC): So, that,

239 00:38:39.810 --> 00:38:55.650 Stephanie Dock (DDOT | DC): there's an opportunity to potentially switch the system over, it's just working off of legacy systems and processes that people are used to and trying to update, so… But as a city in the transportation space, we are trying to kind of consolidate down that we're… all the data comes in the same way from everyone.

240 00:38:57.850 --> 00:39:09.490 Michael Schnuerle (OMF): Yeah, thank you. I think that was part of the sentiment from people at the, at the hearings, this week as well, including the head of the… the director of the for-hire vehicle.

241 00:39:10.110 --> 00:39:14.290 Michael Schnuerle (OMF): group, that it should be consistent.

242 00:39:15.090 --> 00:39:21.700 Michael Schnuerle (OMF): Yeah, I have other questions I could ask, but I'm going to leave it open for other people on the call to ask some questions.

243 00:39:30.520 --> 00:39:47.390 John Clary (City of Austin, TX): This is John again. I'm pretty ignorant about the state of the art with this kind of stuff, and I'm excited to spend some time with the release candidate spec, but I'm curious to know more if anyone can tell me a little bit more about how near misses are,

244 00:39:47.590 --> 00:39:51.020 John Clary (City of Austin, TX): Captured in data, and how those are detected conventionally.

245 00:39:53.140 --> 00:39:55.969 Stephanie Dock (DDOT | DC): I'm just gonna pile on and say I would also like an answer.

246 00:39:56.580 --> 00:40:05.740 Stephanie Dock (DDOT | DC): Near-miss definitions are notably lacking. Not in your book, just, like, in the field as a whole, it's… Yes.

247 00:40:06.080 --> 00:40:22.290 Michael Schnuerle (OMF): So something… John, you may know this, having been involved with MDS. A lot of times, we sort of give something a name, but in some ways, we don't define it completely, because it varies by state law, or jurisdiction, or country. So, for instance, we have near-miss as an option when capturing an incident.

248 00:40:22.300 --> 00:40:34.139 Michael Schnuerle (OMF): But we don't really define it, so we leave it up to the jurisdictions. And this has been a question already from some jurisdictions, like, how do you define it? We define it this way. So, I'd be curious, because DC is…

249 00:40:34.290 --> 00:40:45.410 Michael Schnuerle (OMF): going to be asking for this, and so there has to be some definition somewhere. And then, Robbie, if you're still on the call from FGM, you're getting this from Cowboy, how do… how do they or you define it?

250 00:40:45.860 --> 00:40:48.060 Michael Schnuerle (OMF): Maybe he's eyes off, yeah.

251 00:40:49.860 --> 00:40:51.830 Robbie Vinogradov (City of Amsterdam): Sorry, you were talking about the near misses?

252 00:40:52.050 --> 00:40:56.329 Michael Schnuerle (OMF): Yeah, how do you define in your myths specifically with the cowboy data, do you know?

253 00:40:56.510 --> 00:41:11.430 Robbie Vinogradov (City of Amsterdam): Yeah, so the cargo data only distinguishes between a minor and a major incident, and like I said in the introduction, the minor incidents are, are determined by the feedback that's provided by the rider.

254 00:41:11.800 --> 00:41:28.990 Robbie Vinogradov (City of Amsterdam): So it's basically how the rider perceives the incident, to let Cowboy know whether it was a minor or a major incident. And how it exactly works, how Cowboy determines a major incident is also a bit unclear to me, to be honest.

255 00:41:30.000 --> 00:41:31.849 Michael Schnuerle (OMF): Okay, it's a little fuzzy.

256 00:41:32.300 --> 00:41:32.990 Robbie Vinogradov (City of Amsterdam): Yep.

257 00:41:34.940 --> 00:41:39.589 Stephanie Dock (DDOT | DC): Yeah, I… I know that Near Abyss is in the feed, I don't know that we are…

258 00:41:40.220 --> 00:41:45.349 Stephanie Dock (DDOT | DC): I think we will be leaning in on the non-near-miss portion.

259 00:41:46.730 --> 00:41:48.549 Stephanie Dock (DDOT | DC): Just recognizing that

260 00:41:48.830 --> 00:42:04.380 Stephanie Dock (DDOT | DC): we don't have any thresholds set on that. I think, like a lot of folks, you either have kind of camera or sensor-based detection at intersections that someone is selling you the possibility of identifying near misses, and if you've delved into this space at all, it is…

261 00:42:04.530 --> 00:42:13.570 Stephanie Dock (DDOT | DC): there's a lot of different definitions, and what actually is kind of an actionable insight, right? Like, if a pedestrian walks out into a street.

262 00:42:13.570 --> 00:42:25.890 Stephanie Dock (DDOT | DC): knowing full well that there are cars coming, but they're, like, timing their walk to the speeds of the cars, they will set off a proximity threshold, but you have to be tracing the trajectories to see whether they're actually gonna…

263 00:42:25.890 --> 00:42:39.759 Stephanie Dock (DDOT | DC): interact. And I've mostly been looking at the near-miss conversation from the fixed sensor perspective, and I think there might be some side things on the vehicular side, but harsh braking is often used as a proxy.

264 00:42:39.810 --> 00:42:53.630 Stephanie Dock (DDOT | DC): Because it's… indicates something happened that was an issue. But mostly on the vehicular side, right? Because you… the OBD2 port can report that kind of information. So, sorry. Aspen has their hand up.

265 00:42:54.180 --> 00:42:55.610 Michael Schnuerle (OMF): Yes, that's true.

266 00:42:57.340 --> 00:42:58.060 Espen Johnsson: Hello?

267 00:42:59.130 --> 00:43:00.310 Michael Schnuerle (OMF): Yes, hello, Espin.

268 00:43:00.310 --> 00:43:17.489 Espen Johnsson: Yeah, thank you. Espin here from Oslo Institute of Transport Economics. I've looked a bit at this sensor data for the last, like, 5-6 years, and, at least on the, on the near misses, or…

269 00:43:17.580 --> 00:43:35.000 Espen Johnsson: on air accidents, type of thing. At least what's been published in papers is, it seems like people use a combination of, some experiment with… with the kind of radar, LiDAR-type sensors, but that… that won't scale.

270 00:43:35.190 --> 00:43:44.559 Espen Johnsson: In Oslo, one of the operators have had an AI vision-based solution on their

271 00:43:44.830 --> 00:43:50.339 Espen Johnsson: Well, 20% of the fleet or something, like, a few thousand scooters.

272 00:43:50.610 --> 00:43:56.400 Espen Johnsson: And others use, accelerometer wheel speed, and…

273 00:43:56.540 --> 00:44:03.110 Espen Johnsson: If it's possible to capture it, the, direction of the front wheel.

274 00:44:03.490 --> 00:44:15.029 Espen Johnsson: I get quite a lot of data from the operators, but I haven't seen that any of the scooter operators, at least those I've been in contact with, have… will capture the…

275 00:44:15.260 --> 00:44:24.680 Espen Johnsson: the direction of the front wheel. So, so, but, usually it's… it's somewhere between 5 and…

276 00:44:24.970 --> 00:44:29.779 Espen Johnsson: 12Hz on the sensor side, and 1 or 2Hz on the GPS.

277 00:44:30.350 --> 00:44:35.699 Espen Johnsson: Hmm… But on, just to mention it,

278 00:44:35.990 --> 00:44:43.249 Espen Johnsson: for harsh braking, I've done that quite a lot on that, and it's… You get kind of…

279 00:44:43.490 --> 00:44:54.139 Espen Johnsson: A lot of noise, so if you're gonna do some serious analysis on it, you've got to capture the street… the singles, the single timing.

280 00:44:54.950 --> 00:45:03.600 Espen Johnsson: Because, like, 80% of even the hearts breaking, so… around… Signaled crossings.

281 00:45:04.730 --> 00:45:15.720 Espen Johnsson: And that's, at least here. I haven't been able to trace if anyone actually logs those data, especially on dynamic signaling systems.

282 00:45:15.880 --> 00:45:19.279 Espen Johnsson: It's very few that are very static, so you can see, like…

283 00:45:19.650 --> 00:45:33.989 Espen Johnsson: every 15 seconds as, well, yeah, whatever it is for, for, for, for a minute, but most of them are some kind of dynamics, so there's no… no predefined,

284 00:45:34.200 --> 00:45:37.049 Espen Johnsson: Timing, you can… you can work out from.

285 00:45:38.040 --> 00:45:40.840 Espen Johnsson: And that's quite a big deal.

286 00:45:41.760 --> 00:45:42.290 Espen Johnsson: Okay.

287 00:45:42.290 --> 00:45:48.820 Michael Schnuerle (OMF): Thanks, thanks for sharing. If you have any links to information about that work, feel free to put it in the chat.

288 00:45:49.200 --> 00:45:49.550 Espen Johnsson: Hmm.

289 00:45:49.550 --> 00:46:03.200 Michael Schnuerle (OMF): I will say, something that was hinted at by Stephanie, I think, is that we had talked about providing this sort of information through MDS back in 2020, I think? Maybe 2021.

290 00:46:03.300 --> 00:46:09.869 Michael Schnuerle (OMF): But it really felt like the operators actually felt like they weren't able, as scooter companies, to

291 00:46:09.870 --> 00:46:25.060 Michael Schnuerle (OMF): determine if there was a near miss, or even sometimes if there was actually a crash. But now, as we have, you know, autonomous vehicles and sidewalk robots, I think this information is more… and e-bikes of different… with different sensors, this is more accessible.

292 00:46:25.060 --> 00:46:29.120 Michael Schnuerle (OMF): But many of these terms, probably, and the definitions of them are going to be

293 00:46:29.180 --> 00:46:33.600 Michael Schnuerle (OMF): different by jurisdiction. Does anyone else have thoughts on, near misses?

294 00:46:36.080 --> 00:46:47.249 Michael Schwartz, INRIX: I just wanted to jump in. This is Michael Schwartz from INRIX. We're a long-time consumer of MDS data, but now we're also a provider of MDS data on behalf of one of the robot delivery companies, and

295 00:46:47.250 --> 00:46:57.240 Michael Schwartz, INRIX: I think it's probably just a wide acknowledgement that there is both a science and an art to developing MDS data, and my guess is that will be the same when it comes to incidents.

296 00:46:57.240 --> 00:46:58.959 Michael Schwartz, INRIX: You know.

297 00:46:59.600 --> 00:47:16.469 Michael Schwartz, INRIX: There's part of me that feels like it's impossible because of, sort of, trade secrets and business needs, but it would be great to have some collaboration amongst the operators and providers on how to define these things, because it does seem like each company defines… you know, the way they do operations is different, and how they slot it in, and

298 00:47:16.470 --> 00:47:35.880 Michael Schwartz, INRIX: We've had some interesting back and forths with a number of companies. I see Pierre is on with Blue Systems in LA about just trying to understand how a robot delivery service even operates. It's just really, really different than how a scooter service operates. And obviously autonomous vehicles will be even different than that. And so, it might just be something we want to think about for

299 00:47:35.880 --> 00:47:43.470 Michael Schwartz, INRIX: how to bring some of those providers into the fold, and if there's any ability to have a collaborative forum here, I think that would be really, really helpful.

300 00:47:43.470 --> 00:47:47.330 Michael Schwartz, INRIX: You know, as we start to get feedback on something like incidents.

301 00:47:48.890 --> 00:47:54.950 Michael Schnuerle (OMF): Yeah, thanks for that. I'd say the OMF has done that in the past, and we'd love to do that more, and I know we have

302 00:47:55.140 --> 00:48:07.240 Michael Schnuerle (OMF): more and more operators participating. I see some on the line and some new members to the OMF, so happy to be that convener and discuss some of those details, Stephanie.

303 00:48:09.060 --> 00:48:10.720 Stephanie Dock (DDOT | DC): Just a…

304 00:48:11.080 --> 00:48:26.379 Stephanie Dock (DDOT | DC): thankful that Michael has brought that up, but also, Aileen, you mentioned in the comments that it's remembering what this data is ultimately being used for, and I think it is a bit of a distinction in some ways between the uses on, for example, our permitting for AVs and PDDs.

305 00:48:26.380 --> 00:48:36.009 Stephanie Dock (DDOT | DC): that fundamentally, it's safety, but for us, that's also a critical oversight function in a way that's different from how it goes with bikes and scooters, right? I…

306 00:48:36.010 --> 00:48:47.739 Stephanie Dock (DDOT | DC): I'm much more incentivized to require it, because if there's a problem, we need to be addressing how their operations are occurring in a way that if you're renting to individual users who you don't have a lot of control over.

307 00:48:47.890 --> 00:49:04.279 Stephanie Dock (DDOT | DC): you know, it's that system safety that you're trying to improve, and somebody… you can work on behavioral, right? Whereas, like, the company is operating the robot, and, you know, it's… where it's our context that's causing the problem, absolutely, we need to fix it, but there… I feel like there is… and I think that's…

308 00:49:04.300 --> 00:49:10.370 Stephanie Dock (DDOT | DC): Why some of the conversation… that both the devices are capable of different things, but what we can also expect of them as a regulator.

309 00:49:10.550 --> 00:49:17.599 Stephanie Dock (DDOT | DC): I think change a little bit, where if we're consistently seeing a sidewalk robot, like, running into stuff.

310 00:49:17.600 --> 00:49:32.340 Stephanie Dock (DDOT | DC): We need to talk about what their sensors are detecting and… and what's going on, in a way that, like, if scooters keep running into things, I might want to talk to the people about how they're… how they're running. So it's… it's…

311 00:49:32.570 --> 00:49:46.899 Stephanie Dock (DDOT | DC): I think it is ultimately back to saving lives and preventing serious injury, but some of the steps in the intermediary, I feel like, are different from our sides, and that's why, like, you're hearing from my program, and maybe not the shared fleet device or the car share side.

312 00:49:48.450 --> 00:49:51.210 Michael Schnuerle (OMF): Thanks, Stephanie. Byrne from URF.

313 00:49:52.380 --> 00:49:58.110 Bern Grush, Urban Robotics Foundation: Yes, thank you. I'd just like to point out that there is a… there's work underway, and it's not completed.

314 00:49:58.240 --> 00:50:14.980 Bern Grush, Urban Robotics Foundation: for something under ISO 4448 called a Journey Data Recorder, and it's for what we call public area mobile robots, of which we're talking about PDDs mostly. And that is… that Journey Data Recorder is there for purposes of training.

315 00:50:15.110 --> 00:50:16.470 Bern Grush, Urban Robotics Foundation: Insurance…

316 00:50:16.610 --> 00:50:23.590 Bern Grush, Urban Robotics Foundation: Near misses and crashes, and we have the same problem. We don't have a definition for near miss, we have a definition of

317 00:50:23.850 --> 00:50:33.270 Bern Grush, Urban Robotics Foundation: How close things are, and, you know, certain distances about… certain definitions about distances, but it's not a definition of near or miss.

318 00:50:33.480 --> 00:50:50.080 Bern Grush, Urban Robotics Foundation: In that air… in that work, I think there is some opportunity to establish the definition of an ear miss for those kinds of devices. It's not a general definition, but I just wanted to point it out, and I'm certainly happy to talk to anybody about that, if it's appropriate. Thank you.

319 00:50:50.830 --> 00:50:58.419 Michael Schnuerle (OMF): Thank you, Byrne, for your expertise, and again, feel free to maybe put a link to that in the chat, if you can, about the ISO standard.

320 00:51:01.140 --> 00:51:07.310 Michael Schnuerle (OMF): Any other thoughts or… on this specifically, or in general, for Stephanie?

321 00:51:07.450 --> 00:51:09.250 Michael Schnuerle (OMF): About the work in DC.

322 00:51:17.650 --> 00:51:32.329 Michael Schnuerle (OMF): I'll point out, you know, I mentioned at the top that we know of 200 operators around the world that have devices in the public realm that are using MDS, 200 operators are using MDS.

323 00:51:33.140 --> 00:51:46.209 Michael Schnuerle (OMF): So, you know, Yellow Cab, for instance, we were talking about taxis. Yellow Cab is using it in Los Angeles already, and Yellow Cab came up in DC, so if there was a requirement there, the Yellow Cab has sort of done that work.

324 00:51:46.260 --> 00:51:59.919 Michael Schnuerle (OMF): For LA, and so it should be an easier lift for DC. So, getting the operators together, I think many of them are already using MBS in some capacity, and then this could be… this is a new area for them, if it's about crash reporting.

325 00:52:00.010 --> 00:52:02.290 Michael Schnuerle (OMF): But we do have…

326 00:52:02.490 --> 00:52:09.940 Michael Schnuerle (OMF): you know, a list, and we have reached out, but, I think, maybe a future meeting, like Andrew suggested, could be about

327 00:52:10.400 --> 00:52:16.110 Michael Schnuerle (OMF): operator, definitions and uses of some of the new features of MDS 2.1.

328 00:52:20.950 --> 00:52:22.390 Michael Schnuerle (OMF): Alright, let me,

329 00:52:23.110 --> 00:52:29.050 Michael Schnuerle (OMF): I'll just wrap it up, but if people have final comments, feel free to put it in the chat. Thank you.

330 00:52:29.220 --> 00:52:30.740 Michael Schnuerle (OMF): Burn for that.

331 00:52:31.220 --> 00:52:32.500 Michael Schnuerle (OMF): ISOLink.

332 00:52:37.330 --> 00:52:43.210 Michael Schnuerle (OMF): Yeah, and the wrap-up is just that our next meeting is gonna be July 2nd,

333 00:52:43.210 --> 00:53:06.979 Michael Schnuerle (OMF): It may be pushed to July 9th because of the U.S. holiday on the 4th. We're not sure, but it will probably be around the launch of MDS 2.1, so you'll get more details about MDS there and the final version of it. Stephanie mentioned it's still under review, and it is, in a way, but the board has approved it, so if you look at the development branch of MDS repo right now.

334 00:53:07.300 --> 00:53:21.479 Michael Schnuerle (OMF): Actually, the 2.1 branch is probably more accurate at the moment. That is 99.9% gonna be what the final version of MDS 2.1 will be, and that is on the release plan page that I sent out earlier.

335 00:53:21.550 --> 00:53:40.769 Michael Schnuerle (OMF): So anyway, that… the final thing is just to package it up and provide some launch materials. So we will be working on that and presenting it at the next meeting. Feel free to start to learn and understand 2.1, and even start implementing it, like many cities and companies have already started on.

336 00:53:41.440 --> 00:53:46.640 Michael Schnuerle (OMF): So thank you, here's how you can reach us. Website, LinkedIn, GitHub, and YouTube.

337 00:53:47.230 --> 00:54:10.669 Andrew Glass Hastings (OMF): And Michael, quick reminder, just a plug for Open Mobility Foundation members, the MDS Working Group Steering Committee nominations are out right now. More information on how to nominate yourself or nominate a colleague, coming out in the OMF member newsletter later today. Again, that's for OMF members.

338 00:54:10.670 --> 00:54:33.629 Andrew Glass Hastings (OMF): For… to be a part of the steering committee. If you are interested in that, please reach out to Michael or any of us on the OMF team. If you're not yet a member and interested in getting more involved in helping drive the development and future of MDS, please also reach out to Michael, Aileen, or myself. We'll be happy to talk about what membership looks like.

339 00:54:37.560 --> 00:54:40.270 Michael Schnuerle (OMF): Alright, thank you all, and

340 00:54:40.660 --> 00:54:55.209 Michael Schnuerle (OMF): if you reference the chat, I'll put the chat log in the meeting notes as well, but there is information about, an upcoming Urban Robotics Foundation webinar and, email address there, so thank you for that.

341 00:54:55.410 --> 00:55:01.459 Michael Schnuerle (OMF): All right. Thanks, everyone. Thanks to our presenters today for presenting, thanks for all of our attendees and the great questions.

342 00:55:01.800 --> 00:55:04.339 Michael Schnuerle (OMF): Have a good day, and we'll see you next month.