What is OSM-based car navigation like in Southern California?

Communities United States
1

In the last couple of years I have intensively mapped the Southern California region from imagery (Los Angeles and San Diego metropolitan areas, Imperial Valley). I am from a different part of the world, but my thinking is that if this hugely populated and culturally influential region were mapped well, it would be a big boost to OpenStreetMap as a project.

However, I don’t know how much of a real-world impact all these thousands of changesets by myself and other people have had: turn lanes, speed limits, stop signs, etc. In this region where the car is king, is OpenStreetMap competitive enough now for anyone who wants to navigate with e.g. Organic Maps in their car, or do even OSM enthusiasts in this region still feel the need to use Google-provided solutions?

Any suggestions to improve my own mapping by people who live in SoCal and rely on OSM car navigation would be welcome.

3 Likes
2

This will, of course, not be a scientifically valid poll or sample. :slightly_smiling_face:

I will start with letting you know that I live in Southern California so I guess I am in your target responder pool.

For my self, in the last 6 months to a year became aware of research showing the negative impact on spatial memory from habitual use of GPS apps. So I have switched from using any map app while driving to simply navigating by memory. I have noticed a decided improvement in my grasp of this area that I moved to nearly 9 years ago but never fully learned. Now, if it is not a destination I am familiar with I look up the location, usually on OSM and a suggested route on Organic Maps and commit the significant turn locations to memory. So I might not be the best person to answer your question with respect to using OSM based apps for navigation.

For me, the coverage of stop signs, speed limits, turn restrictions, etc. is nice simply because it can give a more accurate estimate of a good route and typical travel time.

Turn lanes are not quite as important as they don’t typically affect route selection or travel time. Back when I was using map and navigation apps for turn by turn directions it was nice to know in advance what lane(s) I should be in for the next turn. But even then, much less important to me than knowing the route and a predicted time of arrival.

For me, the most important thing is getting POIs and all the address data into OSM. Things are a lot better now than a few years ago, but there are still places that I have to look up using a non-OSM map to find where they are.

Address data is becoming more available in datasets with licensing compatible with OSM. While I still prefer gathering addresses by walking around it is often possible to do that remotely. In my opinion, gathering POIs is best done locally by survey though street level imagery can get you started.

1 Like
3

I grew up in San Diego (in the 1960s and '70s), often visited (family, cultural events…) in Los Angeles (and environs: Orange, Riverside counties…), am very much an automobile-based driver (though, I have taken a lot of public transit in California, both Southern and Northern, too) and have “GPS navigated” since the late 1990s or so with consumer-level Garmin products that I deeply used, filled up their memory (with, for example, “freeway crossroads” waypoints) and bought newer versions of when those came out and allowed me expansion capabilities of GPS, storage of my data and interface-ability with computers + mapping software and other cartographic systems. All of this (and more, like being a computer scientist and employee stints at Apple and Adobe) were a good springboard into OSM when I joined in the late 2000s.

One of the most important tasks I find to be both challenging and rewarding in OSM is to grok the whole area I might work / play / roam in and determine what is missing or incorrect. It might be highways, it might be recreation opportunities (e.g. biking or hiking trails), it might be that POIs need updating, it might be that public transit needs improvement, it might be that admin_level settings kinda suck and need for others in the area to come to some sort of agreement that improvements are due, and then together, with consensus, we make them.

I also have a pretty good “spatial memory” (in my brain) for self-directed navigation and know exactly what you mean when you imply that GPS-directed car navigation can become a “crutch,” causing a sort of “brain atrophy” at one’s ability to self-navigate (without GPS navigation, or even a paper map).

Your (@CRCulver 's) propensity to map “turn lanes, speed limits, stop signs…” especially in Southern California where indeed the “car is king” is yet another case of OSM seeming to absorb what is relevant in various parts of the world: where train stations and passenger rail is relevant (say, Europe or Japan), OSM finds really rich coverage of these in those areas. Around “here,” automobile-oriented data are similarly important. Of course, everywhere POI data need to be both added and updated, and also things like recreational opportunities…people DO use OSM to plan and navigate something like a bike trip, enjoying amenities at a park, even going backpacking (like me).

I’d say “keep up the good work,” and suggest (if you haven’t done so already) to “complete the feedback loop” by using OSM data in your GPS. There are methods to convert OSM data into, for example, Garmin map data, allowing you to install these into your device (such as via a USB cable, Bluetooth or onto an SD card) and then you can navigate (visually or via text-to-speech, if your device supports that) with OSM data itself. This can show you “holes” in our data, which can prompt you to sharpen focus on entire aspects of our data (like off-ramp exit destinations and much more).

I’m just throwing out ideas, but I think you can see the picture I’m painting.

4

grafik

7 Likes
5

Thanks for the responses. Obviously lack of POIs is a problem in SoCal like everywhere else in the USA, but there isn’t much I can do remotely. (There wasn’t much I could do during the three-week visit to SoCal that sparked my initial interest in the region; distances between properties were so vast that it wasn’t as efficient to walk down the street and map all POIs like one would do anywhere else in the world, and once I was stopped by local police and asked questions on the grounds that “nobody walks around here”.)

I don’t doubt that OSM has been useful for getting around for well over a decade, as @SimonPoole notes. What I want to get a sense of is how smooth the experience is, and whether it is competitive with commercial options. Are drivers alerted to turn lanes at major intersections more often than not? Does routing minimize the number of stop signs they are sent through, or are there places where they feel like they are hitting stop signs at every intersection? And so forth.

5 Likes
6

If you haven’t yet, check out some of the existing OSM-based navigation applications and see if they allow you to simulate your location instead of being there in person. If not, there are tools for fooling Android or iOS into thinking you’re somewhere you aren’t. I use these tools frequently when diagnosing bug reports and dissecting claims about how other routers work.

As others have noted, lane-level details usually aren’t very meaningful to locals who already mostly know the area and have a good sense of direction. Most mappers have a well above average sense of direction. :wink: Outside our bubble, lane guidance has long been a very popular feature in navigation systems, so lane counts and turn lanes are likely helping people day to day. Lane-level details give OSM more credibility as a source of navigation data and OSM-based navigation systems more credibility as consumer-ready products.

One tip that might interest you is that some routers only evaluate the turn:lanes=* tag on the very last way before the intersection. A great many intersections in SoCal are between two divided streets forming a :hash:. I’ve often noticed that little ways inside the intersection, such as this little bit of Valley View Street in Cypress, lack turn lane tags compared to preceding ways. This can sometimes foil lane guidance if the router doesn’t realize the :hash: represents a single intersection.

This is not to say that the turn lane’s overall length is unimportant. I spent a lot of time mapping turn lanes in the Bay Area, along even the most minor of streets and even parking lot exits, in order to generate aggregate statistics about turn lane lengths that pretty much no one had ever gathered before. These statistics helped inform the timing of turn instructions in two routing engines (though ideally the routers would handle the lane lengths on a case-by-case basis).

Some routers like Valhalla do penalize stop signs. This is a relatively undeveloped aspect of routing development so far, partly because coverage is so sparse. (Chicken, meet egg. Egg, chicken.) A StreetComplete quest about intersection control would go a long way toward the kind of coverage that would motivate router developers to fine-tune this kind of detail in their profiles.

Stop signs are also useful for guidance. For example, some OSM-based navigation systems mark the locations of stop signs (and traffic lights) along the route line, so the user has a better sense of how much farther to go. In an urban area, a count of blocks ahead is a lot more meaningful than a distance as a fraction of a mile. With better coverage, developers might feel more confident switching voice instructions over to block counts too.

1 Like
7

Hello! I until quite recently lived and mapped in Los Angeles. I’ve definitely seen your name pop up on features and edit histories, had no idea you were from a different part of the world! In fact, I always wondered if your username was somehow related to Culver City.

First off, thanks for contributing to the region! I think every little bit helps, especially in a region as vast as Southern California. I will say as a local that I used multiple mapping softwares, including OSM-based and Google, to get around. OSM’s coverage of things like speed limits were often sparse, or perhaps worse, outdated as roads are reconfigured, so I think keeping things updated is really helpful for increasing parity with commercial apps.

To respond to your questions, I tended to find the biggest drawback of using OSM-based maps for car navigation was its lack of real-time traffic support (at least for the apps I’m familiar with like OSMAnd). Los Angeles has severe and fairly variable traffic, in which the fastest route can vary significantly hour-by-hour and taking the wrong one can lead to tremendous delay. Much of the reason I used navigation apps was to figure out if, e.g. I should get on the freeway or take surface streets at a particular time, even though without traffic one would obviously be faster. Even with correct speed limits mapped, if the router says it’ll take 20 minutes to get from Santa Monica to Downtown LA at 5:30pm it’s not particularly useful, and I wasn’t aware of an OSM app that could do that (though I could be wrong!).

I’ll add though that despite the “nobody walks in LA” refrain (and sorry for your unpleasant encounter!), I also worked to add a lot of the attributes you mentioned, like speed limits and stop signs, for the benefit of pedestrians. If you do dare to walk in Los Angeles, as I often did, it’s really helpful to know e.g. whether there’s a 4-way or a two-way stop at an intersection, or whether this road that has no sidewalks has a speed limit of 25 mph and two lanes or 40 mph and six lanes. Related to this, I tried hard to mark when there were or weren’t sidewalks along major roads and where there might be places to cross the street: this might be something you can do remotely while you investigate the other attributes if you’re so inclined.

But this info is useful for car routers too, especially to have all the info for someone to make a better one. For instance, I always felt that Google insufficiently penalized left turns across major arterials from minor side streets which don’t have the right-of-way: as in, it would often suggest them, when at rush hour such a maneuver is nearly impossible. If there were good coverage of stop signs, you could make a router yourself that adequately penalized such a behavior. Now if only we could map the presence of a left-turn signal…

1 Like
8

Have not been to South California but generally to me I really a hoy having Lane guidance. Especially in cities where you have not been before, know when to get onto which lanes keeps you from having to switch across 3 lanes the very last moment.

@willkmis MagicEarth uses OSM based maps but also offers live traffic, if you want to give it a try. =)

9

Based on what I saw on adjoining streets when I was lane-mapping I 405 about 8 years ago, I’d expect the lane guidance to be considerably worse than in almost anywhere in Oklahoma.

10

I have always tagged these intersections appropriately. However, there was a corporate-mapping project some time ago that added turn lanes to primary highways all over the region, and there the corporate mappers sometimes failed to correctly tag also the inner ways. There may be a lot of cleaning up to do.

Thanks for drawing my attention to the Cypress area, I will be mapping in that area shortly. In general, I have been focusing more on the periphery of Los Angeles, on the (perhaps mistaken) assumption that the core of Los Angeles would be considered so important by corporate actors that they would see to it.

That’s interesting, I didn’t know this is a problem. Bing aerial imagery doesn’t show streets being congested (which is good, because otherwise I wouldn’t be able to see a lot of the markings painted on the road). What do people do when they live in a residential development where there is only a single exit road, and going where people would want to go requires a left turn onto the thoroughfare? Are they stuck at the intersection for long periods at rush hours?

It feels like something is missing from this post. Could you tell us more?

11

The lane mapping in So Cal is somewhere between “non existant” and “was once accurate but the lack of rendering in id has caused people to ignorantly munge it” with a side order of “not many active mappers in the region”.

12

OSMand has a simulation mode where you can set a start point, rather than use the GPS location.

1 Like
13

Likewise, Organic Maps allow you to set the “route from” in addition to the “route to”.

14

@Baloo_Uriza , Eight years ago is a long time when it comes to lane mapping (I and some other mappers I know only really became interested in mapping turn lanes around 2020–2021). I’d encourage you to give the region another look. Whenever I zoom in on some area showing signs of incomplete mapping (e.g. duplicate traffic signals), I find that more often than not, the turn lanes are already there on primary and secondary highways. Sure, the inner ways are sometimes incorrectly tagged, as mentioned above in this thread, but corporate mapping efforts have had a big impact.

Obviously turn-lane coverage on tertiary highways and residential roads is much less complete, though I assume drivers are less stressed about entering the right lane in time on those roads.

15

At least mostly negative in Oklahoma. Actually having to clean up after Tomtom rather extensively today after they kinked and significantly lengthened every offramp in Oklahoma for no reason or basis in ground truth last night.

1 Like
16

If you’re referring to the practice of moving the start of the ramp to the start of the deceleration lane, I also experienced similar edits by Telenav in the Bay Area several years ago. The good news is that they haven’t come back and engaged in an edit war after I fixed them and added change:lanes=* tags. Lyft subsequently made it impossible for them to go back to that style by adding destination:lanes=* everywhere, probably more aggressively than I would’ve.

I have witnessed or engaged in minor edit wars with other navigation mapping teams over one-way streets in my area, sometimes between two different teams working from their own imagery of the same street, but they usually back off after a few attempts and a long note=* and lots of micromapping.

17

In most cases, there are no acceleration or deceleration lanes in Oklahoma (yield signs and stop signs on onramps exist for good reason: You don’t have the room to shop for a gap after passing it), so they’re mostly just extending arbitrarily and adding a sharp kink next to the gore.

18

Ah, the OSM community dynamic in one paragraph.

1 Like