The Vegas Loop just expanded to four more stations and even uses surface roads for some of them. Right now, the relation is a big mess, and I’m not even sure if the member ways are in the correct order. (That’s how you know you’re doing public transit correctly! /s) I have several questions about the relation and its members.
I tagged the relation and its member ways with ref=Loop. I know this isn’t a highway, but I felt like adding a ref tag to the relation’s member ways will help users know what highways make up the Vegas Loop. Is there a better way to do this? If so, what is it?
On a related note, how could I get the Vegas Loop rendered with its symbol rendered, say, in OSM Americana?
What is the correct order of the relation’s ways? Right now, I have it tagged as public_transport:version=1 out of convenience. Similarly, what is the correct order of its stations?
How should the stations themselves be tagged? Right now, amenity=taxi is the “default,” and I guess it kinda makes sense (except for share_taxi=yes), but is there an even better tag?
OSM Americana itself only shows shields along designated highway routes and scenic routes. We plan to branch out to recreational route shields, as well as agency logos on train stations and bus stops, based on the precedent set by transportation atlases. But Americana isn’t a public transportation map, so I think it’s pretty unlikely that the style would ever use icons for public transportation routes along those roads, let alone an imitation of public transportation.
If you like the idea of iconifying the stops, consider getting an entry for the Vegas Loop into name-suggestion-index once you’re sure about the tags for one of the stops. Americana isn’t quite ready for brand icons yet, but that would be the way to think about this contraption, rather than by analogy to a conventional highway route.
Arguably, there should not be a type=route relation, as there is no fixed route. A relation would only be useful for collecting all the utilised roads and ‘stations’ together.
I still stand by my suggestion to use highway=service for the purpose built roads/tunnels that the cars use. They are more akin to dedicated bus roads than an unclassified/minor highway.
Not route==share_taxi / vehicles modes , but might still be type=route akin to a route==railway / =tracks focusing on the roads. Although, there’s no route=busway for BRT and other buses to compare with yet.
Slightly relatedly, scalability is the reason I don’t prefer highway=busway , as highway=taxiway (???) would be needed for a dedicated exclusive closed-system autonomous vehicle roads for taxis. Another case is a =truckway for a company’s 30km (may be short to other countries) intercity 2+2-lane divided grade-separated controlled-access high-speed road. Category:Ube Industries Ube-Mine Expressway - Wikimedia Commons
Can you elaborate your reason? Of course, the used “buses” are on the small side and the routes are more flexible, but there are dedicated stations. I think that tagging these purpuse-built roads and tunnels as specified on Tag:highway=busway - OpenStreetMap Wiki looks like a good idea. (If the technology becomes more robust and widely available, I would rather expect most existing busway systems to use autonomous vehicles - and I don’t think it would be necessary to change the tagging of the used infrastructure.)
Why invent completely new tags, if it can be adequately tagged using existing tags? Isn’t this a highway=trunk (or similar, as appropriate) with access=no, and possibly hgv=private ?
The argument would be similar to what’s mentioned by others above “for the purpose built roads/tunnels that the cars use. They are more akin to dedicated bus roads than an unclassified/minor highway.” (although dedicated taxis, not public’s cars). If =busway has been voted to replace =service / grade-separated =motorway / =unclassified , this wouldn’t use =service by the same logic and standards.
But anyway, we could simply ignore this as test/experimental, or gimmick/attractions, thus leaving it at =service
Because the Vegas Loop can’t be best described by the current tagging scheme. The fact that @LordGarySugar and I disagree on what highway=* value the main roads should be is evidence that the current tagging scheme is insufficient to describe the Vegas Loop.
I should mention that the OSM Wiki says that highway=service is used for access roads. The Vegas Loop (the “main part,” anyway) does not consist of access roads. (However, the ones that directly access the stations are.)
“Access road” isn’t the only case =service is being used. That’s wiki being incomplete. It could be used for any internal road. It has been used for many different site-specific roads.
highway=service is a bit of a grab bag. The main characteristic is that it isn’t part of the public street network. It shares this characteristic with highway=track. (This is orthogonal to public access. The public street network facilitates the public’s movement, whereas service roads generally serve specific properties. The highway=track documentation refers to this concept as the “general-purpose road network”.) The individual service=* tags have slightly more concrete definitions, but even those are somewhat determined by intuition.
The American Center for Mobility (“Planet M”) and Mcity are two autonomous vehicle proving grounds in Michigan that have seen a lot of back and forth over the years. ACM features a well-preserved freeway (formerly U.S. Highway 12) that has been repurposed into a test track.
Since highway=motorway is largely defined by physical characteristics in the U.S., the ACM Highway Loop would technically still qualify as highway=motorway and its on- and off-ramps as highway=motorway_link. On the other hand, they would also qualify as highway=service since they’re part of a closed circuit. As a compromise for the renderer, we currently have them all tagged as highway=unclassified, without distinguishing between the main road and its ramps. Curiously, highway=motorway_junction nodes have remained this whole time.
From =motorway to =unclassified , there’s still the presumption of being an external “public road”/“street”/“highway”. And there should be a minimal function of connecting different locations (big, or small inside cities), not inside the same piece of land.
However, =motorway_junction could remain. It means numbered exits, not only “motorway”, or full interchange “junction”. Although, the “public’s road” requirement might need to be added to it, supporting its removal.
Yes, we could arbitrarily require a highway=motorway_junction to occur along a public highway, but it would beg the question: what is Exit 1B? The only alternative I can think of would be to shunt everything about these proving grounds into the nascent mimics:*=* namespace, similar to what we’ve done with xmas:*=* for some seasonal features, lest a router or renderer ignore the blazingly obvious access=private. After that, we can retag the train station at the zoo and the corporate shuttle stop as mimics:public_transport=platform. I’m not sure this is all necessary though. We already have other tools for clarifying the context.
Returning to the closed circuit that started this discussion, highway=service is the least surprising option for the Vegas Loop among the existing highway classifications. The nice thing about highway=service is that service=* lets us clarify the roadway’s function. access=prívate additionally prevents misunderstanding. In the future, we could “graduate” it to a distinct highway=* classification if there’s enough demand to justify the disruption to data consumers.
I’d suggest that usage=tourism might be appropriate (sadly not present there yet). I don’t know enough about the Vegas Loop to know whether that tag would work there too (in addition to whatever else fits); my recollection of Vegas is pretty much that the whole town is usage=tourism
The closest analogy I can think of to this is a people mover system like the one at Heathrow or West Virginia University.[1] Both of those have relatively small pods with low per pod capacity (not quite as bad as the Loop, but approaching it) and a single destination per pod.
Heathrow’s is mapped as highway=bus_guideway with no route relation and the WVU one is mapped as highway=bus_guideway with the routes as route=monorail.
What level of physical interaction do we require for a guideway in OSM? Does it have to rely on physical contact or it it sufficient that a vehicle absent special features would not be expected to manage it at speed without a “bad result”.
Should we have a dedicated route=gadgetbahn tag?
Yes, I am mostly remembering a Tom Scott video. Why do you ask? ↩︎
Despite the marketing, the Vegas Loop still relies on human drivers, not unlike a corporate campus shuttle or a BRT service along a dedicated busway. The technology consists of a closed street circuit, Teslas straight off the lot, and lots of purple and green LED lights. There’s no guideway, just painted edge lines. Future plans for automated ponies and rainbows may scramble our tagging options, but for now, highway=service does a pretty decent job of describing the infrastructure.
From watching further videos on the Heathrow one it looks like it’s only using the kerbs as targets for two pairs of laser rangefinders, so it’s doesn’t seem that unreasonable to stick a system that requires lane assist/ autosteer in the same category (if that’s what the loop is doing, they’re very vague).
I would define a =bus_guideway as guided by physically touching contact (otherwise it’s a narrow road assisted by laser/lidar, similar to optically or magnetically guided buses), and the buses are able to travel off guideway (otherwise it’s a rubber-tyred metro). Ultra fails both criteria. Loop fails the former.
Morgantown was inconsistently (route= unedited, edited by others, and reedited backed) changed from railway=monorail to highway=bus_guideway some years ago unresolved. On the other hand, assuming =monorail must be rubber-tyred as an opposite steel-on-steel =light_rail / =subway doesn’t stand either.=light_rail is what I would used. Changeset: 126707946 | OpenStreetMap
Indeed, Ultra was only further changed from =light_rail to =bus_guideway with a poor reason Changeset: 156944169 | OpenStreetMap