Worker Presence Subgroup meeting, 2022 06 29 - usdot-jpo-ode/wzdx GitHub Wiki

Meeting Information

Worker Presence Subgroup meeting
Date: June 29, 2022
Location: Virtual

Objectives

• Share lessons learned from first data consumer interviews.
• Discuss how to implement new worker presence object in WZDx feeds.

Agenda

• Sign-in and Welcome
• Data Consumer Interviews
• Takeaways for Data Producers
• Discussion
• Action Items and Next Steps

Materials

• Meeting presentation

Minutes

Data Consumer Interviews

Motivation for conducting interviews:

• Validate if the current specification meets the market needs
• Inform a one-pager on the benefits of worker presence information in WZDx feeds

Participants can be characterized under 4 categories:

• ADS Developers: GM, Ford, Kodiak Robotics, Argo AI, Locomation
• Mapping and Navigation: Waze, HERE, Google
• ATMS providers: Q-Free, one.network, Panasonic, GEWI
• Other: Laborers' Fund

General high level of interest in Work Zone data.

• Mapping: Driver alerts and routing
• ADS developers: Help with driver assist
• ATMS: Info for traffic operators; automate some tasks (speed reduction)

Most valuable work zone information:

• Location (roadway, start point, end point, and direction of travel affected)
• Time (start time, end time, etc) were seen as higher priority than real-time worker presence information
  • Real-time worker presence not as high a priority

Uses for worker presence data:

• Proxy for “active” work zone
• Improve routing in navigation apps
• Improve public confidence and awareness
• Both mapping companies and ADS developers expressed concern about incorrect information being passed to drivers

There’s not much interest in a single, specific definition of worker presence. Consumers want information that passes the eye test (i.e. a driver would see workers)

Data quality and accuracy is key for trust and adoption, but most data consumers stated that partial or imperfect data is better than no data

• False positives would decrease confidence in the data
• Much less concern about false negatives b/c almost no worker presence info is available today
• Most data consumers would need information to arrive within minutes

Takeaways for Data Producers

• Accuracy is important, but perfection is not needed
• Timeliness on the order of minutes
• Accuracy issues are more manageable if failure modes are known in advance
• Limited concern around worker presence definition
  • Will drivers “know it when they see it”?

Discussion

Neil: I had the opportunity to participate in the CMV peer exchange and gave an update on what all the (WZDx) groups were covering. One of the rabbit holes we went down was why we aren't using geofencing. I responded that it's come up at different times throughout the process. Their point is that geofencing to identify where work is happening and workers are, vs. wearing a vest (if they leave the site). Geofencing opens up concepts we haven't gone down. Our whole goal is accurate real-time information from the field that impedes the roadway.

Qassim: When looking at autonomous vehicle context, takeaways from producers ("accuracy is important, perfection not required" and "timeliness on order of minutes") become dangerous. Perfection is not important with drivers, but with autonomous vehicles we would need to reach perfection to prepare for AV driving. Same for timeliness, if we have a driver and they can judge, then that's useful. But with autonomous driving, we can't tolerate minutes.

• Kristin: Another comment we heard from the ADS groups that they would never just use a WZDx feed to navigate a work zone. Not a comment on the safety of the ADS. I agree that we want everything to be perfect, but the accuracy and timeliness that ADS are holding themselves to is slightly different than what work zone information would be held to. There's a lack of general information overall, so it's important to keep taking those steps rather than say we can't do this.

Sinclair: It's not terribly surprising that they're more focused on time and location of work zone. I think I mentioned a couple months ago that we're working on worker safety. We've gotten technology focused on worker presence and are using it as a trigger for active work zones - that's probably more what we'll end up doing in that case.

• Kristin: I hear that you'd be using worker presence to update whether work is actually happening. Any other methods?
• Sinclair: Smart arrow boards is our main one, and all projects let since last fall are required to use arrowboards. Those will be integrated into ATMS and will be triggers. Smart rumble strips, AVL data from maintenance vehicles, etc
• Kristin: We're doing similar work in Minnesota with maintenance vehicles

Adam: Some legal issues of how things are defined. Arizona has a double fines statute with a clear definition - that could be useful to tell drivers that fines are doubled.

• Adam: From experience working with developers, we need a live WZDx API but it's a work zone and there's so much fluidity. Data is never going to be 100% accurate. We use and find value in the feed, and it's helpful to others for redundancy.
• Adam: When it comes down to it, they need information on lane closures. They can be mapped by progressive vehicles (i.e. not mapped ahead of time, necessarily). Mostly, when they need to merge. Defining work zones as more impacting lanes, that's what drives that clientele. I need to make this an issue for them because people get run over but there is value to it and we need to make that clear to people.
• Kristin: What comes first is how we build these things out and starting with a bare minimum.
• Adam: I'm talking to people in the safety area, so maybe I'm being the devil's advocate here. Communication of why this is important to the outsiders is why we have this subgroup.

Maaza: I think it should at least be during the duration of the work and time limits would be useful even though even during off-hours it is still useful to know that it is a workzone.

Ross: I'd be interested to get a consensus from agencies about what tools will be used to determine that workers are present. Arrowboards are a great start. Work trucks in the area, personal devices, vests, are great indicators. This group could go to agencies to learn what equipment they'd use. As we try to push contractors on getting the necessary equipment, we need that information and some agreement. Maybe that's a good next step for this group?

• Serge: Thanks - one thing in place today is the WP element has provision to map what devices are the source of worker presence data.
• Kristin: Consumers weren't interested in the method, but were in the timestamp.
• Ross: But it is important to contractors, so that they know where to put the emphasis. That would be valuable at least from our experiences.
• Kristin: My thinking is that there's not one answer, but room for exploration. In the survey we asked how people would know now - there were a lot of different answers, and I'm not sure there will ever be just one. Maybe we can find the perfect way eventually, but we're not there yet.
• Serge: Radioing in was the most popular way in the survey, but that has flaws and delay. Each DOT will have flexibility to define what is there preferred method. Radio based, app, full automation with equipment? We designed the specification to be flexible.

Eli: Having somebody radio or call in is very 1990s - workers will forget, or do their call in but not get a response and walk away. Person on receiving end then has another job to put it into a database that can announce the call. I'm not saying we can or need to stricly look at automated methodology, but a goal here is to have someone in the field pull a trigger that flips the switch in a database and indicate that work is happening. The other version would be that the signboard can automatically send that when it's activated.

• Serge: The standard provides that flexibility now, with some delay. The DeviceFeed is a way of sending that information - flexibility is there, and down the road we should move away from radio. It's a matter of time and people will get there.
• Adam: This work group is here to look at worker protection. There are utility workers who don't have positive protection. Can we start with that immediate need, then tackle more state by state?

Maaza: Active monitoring is always a need to have a human intervention even in cases of smartboards as in false reporting and malfunctioning equipment.

• Kristin: We heard data consumers want to know how the data could go wrong. Methods capture sources, but that could be extended to indicate that there is some limitation to the machine.
• Maaza: It's always better to get some information than no information, even if it's false it's better to be ready for anything.
• Serge: We also heard from mapping and ATMS vendors that it's one source of information among other sources. WZDx would help improve quality of output, and can improve confidence.
• Kristin: Finding errors in your own system is important, and sometimes easiest to understand within a system when there's a problem. That's just as important as all the cool features. Graceful degredation of data is important and often overlooked in race to ship new technology.

Participants

Name	Organization
Donna Clark	American Traffic Safety Services Association
Adam Carreon	Arizona DOT
Jackie Beckwith	Association of Uncrewed Vehicle Systems International
Luke Urie*	Austin Transportation Department
Ben Fischer	Butterfly Junction Technologies
Kristin Virshbo*	Castle Rock
Benjamin Acimovic*	Colorado DOT
David Craig^	General Motors
Eli Sherer	GEWI
Maaza Mekuria	Hawaii DOT
Weimin Huang	HERE
Todd Hartnett	Hill and Smith
Pete Krikelis	Hill and Smith
Jacob Brady	IBI Group
Michelle Boucher	IBI Group
Ross Sheckler	iCone
Sinclair Stolle	Iowa DOT
Alexander Lemka	Maricopa County DOT
Neil Boudreau	Massachusetts DOT
Michelle Moser	Minnesota DOT
Christopher Poe	Mixon Hill
Tony English	Neaera
Mark Mockett	USDOT Volpe Center
Molly Behan	USDOT Volpe Center
Nate Deshmukh Towery^	USDOT Volpe Center
Serge Beaudry*	Ver-Mac
Erin Schwark	Wisconsin DOT
Qassim Abdullah	Woolpert

* Co-Chair of Spec Update Subgroup
^ Co-Chair of Work Zone Data Working Group