topic Re: Errors in Google, Bing, Esri, etc Basemaps in Roads and Highways User Group (RHUG) Questions

Errors in Google, Bing, Esri, etc Basemaps

ErinLesh1 — Wed, 19 Jun 2019 13:07:30 GMT

I'm looking to hear thoughts and experiences regarding updates to incorrect data in popular basemaps like Google, Bing, Esri, etc. Has anyone had success getting a company to update incorrect data in a timely manner? Should government agencies create competing products that are based on authoritative data?

Transportation

Re: Errors in Google, Bing, Esri, etc Basemaps

deleted-user-UxcAu2PHrDQp — Wed, 19 Jun 2019 13:48:21 GMT

I don't worry too much about it. Free is free. We have some relationships with companies and they can harvest our network from open data. We recently changed our creative commons license to make it easier for third parties to grab DOT data. There is money to be made from a road network by private companies, that is their driving factor. I feel they can do it better because things like google navigation, etc are what people are using.

We should complement them by providing the best and most current information possible.Stick to our swim lane and do it well.

Re: Errors in Google, Bing, Esri, etc Basemaps

ScottFierro2 — Wed, 19 Jun 2019 13:54:17 GMT

I've had luck getting Google updated both here in the U.S. and in Africa. While I loathe the workflow and the lack of ability to send them bulk updates the work gets done. U.S. updates are faster but have noticed over time its gone from 1 week to about 48 hours. Unsure if it's because of efficiencies on their end or if they use internal rating for validity of users submitted edits so over time I get less scrutiny at review because I have a 5 star rating vs a 2 star rating. Frustrations can arise due to the workflow because I've had them make updates before that are partially correct (i.e. carry a route name wrong direction at an intersection) and then have to submit corrections which can extend the response time from 48 to 96 hours. Never attempted getting Bing updated nor have I tried getting ESRI updated but ESRI you might see faster results working via a proxy (i.e. get data to DeLorme or NAVTEQ that ESRI pulls in) as opposed to their community maps workflow.

Re: Errors in Google, Bing, Esri, etc Basemaps

ErinLesh1 — Wed, 19 Jun 2019 15:44:26 GMT

What mechanism did you use to work with Google?

Re: Errors in Google, Bing, Esri, etc Basemaps

ErinLesh1 — Wed, 19 Jun 2019 15:47:17 GMT

Thanks Eric. At the DOT, we get contacted by the public to fix issues with street data in private company datasets. I'm seeing a need that isn't being met by these companies and exploring whether government can do something more, although I agree that we provide our best data for free and they can use it as they wish.

Re: Errors in Google, Bing, Esri, etc Basemaps

deleted-user-UxcAu2PHrDQp — Wed, 19 Jun 2019 15:52:29 GMT

Are most of the requests from other government entities?

Re: Errors in Google, Bing, Esri, etc Basemaps

ErinLesh1 — Wed, 19 Jun 2019 16:18:41 GMT

Other gov and the public.... We just had a county Chamber send a letter to the Sec of DOT and Executive Direction of the Turnpike Authority about business owners being upset that their businesses were suffering because several newly constructed roads as part of a new bypass project were not showing on Google.

Re: Errors in Google, Bing, Esri, etc Basemaps

ScottFierro2 — Wed, 19 Jun 2019 16:29:56 GMT

Unfortunately, the only interaction to date they have told me they allow is via registered users within their Google Maps application. Sadly, it was easier to get licensing approvals for use of Google Maps on a television broadcast.

Why I mentioned before I loathe it. Essentially, login to Google Maps > click the menu on left side > toward bottom click Send Feedback > select the correct option for "Missing Place", "Missing Road" or "Wrong Information" and follow the workflow. It's a single edit workflow so you will rinse/repeat this a troubling amount of times for each edit you want to make. Additionally, it's a point driven workflow so if you need to adjust a route or add a route you will identify a single point by clicking on the map and then have to spell out the linear correction in the comments of the edit. This is why it happens with the partial errors that you need to re-identify the location and re-do the comments to detail what they got wrong from your initial submission but to date can say I've never had to make more than 2 submissions. As a final point, the only upside to this I have found is the workflow forces you to see their maps/imagery and in cases where you are adding a missing route I can say that if it's not in their imagery it becomes a challenge and I've found it's easier to wait until the imagery shows the new road before submitting the correction.

I've spoken with several people at Google over at least the last 8 years and they all say there's no bulk update option (i.e. send them a KMZ or something OGC compliant, etc.) and this is the only current prescribed workflow for getting this done. Rough but necessary in some cases.

Re: Errors in Google, Bing, Esri, etc Basemaps

deleted-user-UxcAu2PHrDQp — Wed, 19 Jun 2019 16:55:18 GMT

On the money. This is the workflow given to us also.

Re: Errors in Google, Bing, Esri, etc Basemaps

KyleGonterwitz — Wed, 19 Jun 2019 18:48:20 GMT

Bing maps shows a "Prairie Village" neighborhood in Butler County labeled more prominently than the City of Andover. Those from Kansas City know Prairie Village is a suburban city in Johnson County, KS. We have never figured out why Bing maps labels Prairie Village in Butler County, and after many years of contacting Bing about the issue and using Bing for our Basemaps - it's still there, and we now prefer not to use Bing basemaps.

I use OpenStreetMaps for many basemapping purposes, which I like because it is usually updated before anything else, and if it is not I can update it myself including changing of the symbol classification. I can also grab vector data for the State from https://www.geofabrik.de or similar sites and query, overlay, and show vertical bridge relationships between over/under highways in interchanges, all it costs me is attribution "...CC-by-SA" and/or "© OpenStreetMap contributors..." credits on the maps produced.

For Garmin/HERE we happen to have a HERE representative who lives near me, I provide him planning resolution information regularly and he is very competitive about processing it through as quickly as possible. Even though he may have the HERE Maps updated on the day we go open to traffic, it may take 6 months for Garmin to process the updates into their Vehicular GPS device basemaps.

I would say that based on some conference sessions I've been to lately involving transportation technology, connected and autonomous vehicles and all that - there is room for improvement in this area. Silicon Valley tech companies may think they can do everything without DOT participation but the fact is they can't. They have to admit they have a problem before we can work with them to solve it, I think some states have solutions ready once the problem is recognized.

I wish I could link to some of the specific conference presentations I've seen in the last year. Here are some articles I just googled on the subject:

https://gis.usc.edu/blog/self-driving-cars-and-the-role-of-gis-in-future-transportation/

https://www.gislounge.com/spatial-challenges-navigating-rural-roads-self-driving-cars/

https://spectrum.ieee.org/cars-that-think/transportation/self-driving/mit-experts-selfdriving-cars-wont-need-accurate-digital-maps

I noticed these articles focus on silicon valley technology and never mention state DOT's.

Re: Errors in Google, Bing, Esri, etc Basemaps

Anonymous User — Wed, 19 Jun 2019 19:20:26 GMT

Esri provides a mechanism to submit updates to basemaps here: http://communitymaps.arcgis.com/. There are several tools depending on what you want to provide and the basemap that you want to see updated.

Re: Errors in Google, Bing, Esri, etc Basemaps

AndrewVitale3 — Mon, 24 Jun 2019 16:19:23 GMT

What about the other way around?

Has anyone thought much about leveraging these products to identify locations in your LRS that need to be updated? I think OpenStreetMap would be the best candidate for something like that.

Re: Errors in Google, Bing, Esri, etc Basemaps

NicoleHanson1 — Mon, 24 Jun 2019 16:22:50 GMT

We use basemaps to fill in our LRS. There’s still quite a bit of the public rural roads we don’t have – and didn’t get the first time around with the ARNOLD request.

Re: Errors in Google, Bing, Esri, etc Basemaps

ScottFierro2 — Mon, 24 Jun 2019 16:28:18 GMT

Like this direction but my immediate frustration would go back to some of the update workflows. I almost foresee some fun ML/AI workflow run routinely to calc differences where you look for change (date myself with imagery analysis here but red = fled and blue = new) between LRS and the various sources then submit updates for both new and retired data. The inverse could be done at various upper echelons to assist say State identifying new/retired county, township, etc. routes based off what's found in an OpenStreetMap, etc. and not yet pushed up or been digitized into state's data.

Re: Errors in Google, Bing, Esri, etc Basemaps

AndrewVitale3 — Mon, 24 Jun 2019 16:29:49 GMT

What's the workflow look like? Is it fix it as needed while panning the map with a basemap turned on, or do you guys have a systematic way to go about it?

I would like to figure out if it's feasible to leverage OpenStreetMap data to generate a report showing where OSM changed,and which ROUTE_IDs may need to change with it.

Re: Errors in Google, Bing, Esri, etc Basemaps

AndrewVitale3 — Mon, 24 Jun 2019 16:33:26 GMT

This is pretty much what I'm imagining! Like you said, it would be a fun project. I also think it could be very useful. Many of our updates are currently based on out-of-date orthoimagery. This is getting better as we mature, but the more data sources the better...

Re: Errors in Google, Bing, Esri, etc Basemaps

NicoleHanson1 — Mon, 24 Jun 2019 16:35:27 GMT

Andrew-

It’s fix as needed as we pan around/doing other event edits. We utilize the redline feature from R&H so that any of the business units who enter event data can redline a road that should be on the system but isn’t currently. We have about 5 or so different business units editing in RCE so they each have the potential for finding roads that we might not have grabbed in the first go around.

Also, the office of highway safety uses basemaps to verify crash locations and run into roads that should be on the network, or should be one way but is digitized another. They send us a snapshot of that location and the LRS team fixes it. They are not using the redline feature.

And we’re using basemaps to fill in data for the road names – as not every local agency sent us that data.

Re: Errors in Google, Bing, Esri, etc Basemaps

ScottFierro2 — Mon, 24 Jun 2019 16:42:38 GMT

I know TensorFlow has been used for some imagery classification work but unsure if it's the best for road network extraction. Have seen examples of road network extraction before but wonder if any of the API's would allow for such a workflow to run because often those basemap services don't allow much in the way of advanced processes. Having to do some sort of image capture and then ortho the images yourself first would be tedious and probably more cumbersome than other manual options. Sorry got the gears going now and already thinking out the options and paths to try and undertake this.

Re: Errors in Google, Bing, Esri, etc Basemaps

AndrewVitale3 — Mon, 24 Jun 2019 17:29:09 GMT

Feature extraction from imagery would be awesome... I've kind of resigned myself to that being a pipe-dream at this point. But hey, someone is bound to figure it out.

While at GIS-T, I saw a quick presentation about SharedStreets. Their software is capable of generating a unique ID per roadway segment that will be consistent across datasets. Here's the basic concept:

They also offer OpenStreetMap data via their API, which apparently has already been assigned the SharedStreets IDs. I'm wondering if this could be leveraged to create a report of potential update locations on the LRS.

If you were to split your LRS network at each intersection and generate SharedStreets IDs, you could create something like a centerline sequence table that tracks which SharedStreets IDs belongs to which RouteId. I'm not super familiar with it, but I think OSM data has the last edit date for each segment associated with it. You could relatively easily query for all of the OSM updates within the last month, query the SharedStreets sequence table for the related RouteIds (and m-values if you'd like), then generate a simple PNG of the LRS overlaid on OSM to easily see what's changed.

There's still a lot of thinks and maybes in there, but if we could make something like this work, it would be awesome.

Re: Errors in Google, Bing, Esri, etc Basemaps

ScottFierro2 — Mon, 24 Jun 2019 17:39:24 GMT

Naw, mo pipedream it's happening and coming. I can say back in 2011 we worked with a defense contractor called SAIC and had already been doing automated feature extraction then and were busy trying to resolve issues like dividing individual dwellings from clusters of mud huts and lean-to's. Recent ESRI article actually hits on a road network example.

https://www.esri.com/about/newsroom/arcwatch/where-deep-learning-meets-gis/?adumkts=branding&aduc=email&adum=list&aduSF=newsletter&utm_Source=email&aduca=arcwatch&adut=301117_ArcWatch_June_2019

The SharedStreet's ID is interesting and almost mirrors concepts being used to address long-standing headaches with identity management in enterprise systems.