google utility pole service - usa
TRANSCRIPT
Utility Pole Service
About the serviceExampleHow it worksPole discoverySetting expectationsMid-term plansLong-term plansLimitationsTeamContact
Do you do OSP engineering? Do you wish you could wavea wand and know where every utility pole was in your area? Wouldn’t it be great to have detailed pole metadata available at the click of a button - the data that you need to plan your network - high-accuracy pole location, pole height, conductor altitudes, pole material, and whether there’s a transformer attached? We may have the answer for you!
The Google Utility Pole Service is a new, confidential, utility pole dataset sourced entirely by Google. We are happy to discuss commercial terms for access to this data.
We can deliver in a wide array of formats (SHP, CSV, KML/KMZ) and for multiple compatible systems (ESRI ArcGIS, Ericsson Network Engineer, GE SmallWorld). Pricing is tailored to your needs, such as per pole, km, km² and contracts for twice-yearly refreshes of the dataset.
An example of utility poles provided by our service in a customer-defined area
How It WorksThe utility pole service uses Google’s Street View and aerial imagery, combined with Google-developed machine learning techniques, to (a) locate utility poles, (b) identify a given pole in photographs Google has taken, and then (c) identify particular features of a utility pole, such as its material (wood, steel, or concrete). The data is then manually curated and QA’ed before delivery to a customer.
Why Google Is Providing This ServiceGoogle has the best combination of extensive imagery and machine learning capabilities in the world. This has let us create a useful dataset of indexed aerial infrastructure that nobody else could easily build. After developing this technology and seeing the tremendous impact it has had on our own infrastructure projects, we’ve decided to make it commercially available outside Google. We hope this data will help our customers build quality Internet infrastructure faster to help get everyone connected to abundant Internet. That’s our mission.
Product StatusAs of June 2016, this service has located more than 30 million utility poles in the United States, 5 million in Canada, and one million in Mexico. It is almost certainly the largest database of utility pole locations and photographs in the world. We are expecting the size of our pole database to grow considerably as we integrate additional imagery and refine our processing techniques. The primary product today is a dataset under license from Google containing latitude and longitude of a utility pole with hyperlinks to corresponding Street View imagery. A significant engineering team is working on enriching this database with more information about each pole, including pole heights, existence of a transformer, pole material, and more.
UsesThe service is intended to be useful in both “greenfield” (new build) and “brownfield” (existing build) scenarios and for both pole owners as well as those intending to attach to utility poles.
GreenfieldIf a customer is interested in acquiring new utility pole data in an area where they do not have any current utility pole knowledge, the requestor will provide Google with a polygon outline in KML format of the area with any (e.g. freshness) requirements for the data. Google will check freshness and Street View drive completeness of the area, communicate redrive timing as appropriate, and curate the utility poles within the given boundary. We will then deliver the pole data to the customer electronically in KML, Shapefile (.SHP), or CSV formats as-needed. We can also discuss providing data in other formats like DXF.
BrownfieldIf a customer is already present in a geography and is looking to (a) enrich their utility pole database with poles they may have missed, (b) refine location accuracy, and/or (c) add new metadata such as pole height to their database, we offer a “brownfield data enrichment” service. The customer will provide the KML polygon outline of the bounds for the pole data collection, along with points in the KML file for each currently-known pole in the bounds, and the customer’s internal unique ID for that pole. A Shapefile (.SHP + .PRJ) may also be provided in lieu of a KML/KMZ file. Google will then attempt to match the customer-provided poles to Google-known poles, and will return a three-part KML: the first part will include “matched” poles, where we are confident that we know about the pole located by the customer - we will include our understanding of the pole’s location and associated metadata. The second part will include “Google-missed” poles where we were unable to locate the pole provided to us and thus cannot currently provide metadata for that pole. The third part will include “customer-missed” poles, where we found poles that did not seem to map to poles located by the customer; we will include the metadata known about these poles. (Note that a “greenfield” deploy is essentially this process, but where all of the returned poles are “customer-missed,” since the customer provided no poles.)
Detailed heatmap showing the location of 30 million autodetected utility poles in the USA
Setting ExpectationsBelow are rough expectations about the product we will deliver; note that your executed contract supersedes any numbers given here. In urban areas, we generally expect about 12 linear kilometers of road per square kilometer and about 600 poles per square kilometer for a geography where all infrastructure is aerial (no underground builds). If a selected area differs significantly from this, our delivery times could change. We can scale up our curation & QA process to over 100,000 curated utility poles delivered per month if there is a sufficient need.
1. Turnaround TimeIf no new Street View or aerial imagery collects are required for an area, we expect it to take about 2-4 weeks from receipt of the KML bounds to delivery of the utility pole data.
2. Location AccuracyWe expect that 95% of our delivered utility poles will be correct to within 2m of their true location in areas where we have sufficient aerial imagery quality. See limitations below.
3. Precision & RecallWe expect to have 95% precision and 95% recall for front easement utility poles in areas driveable by Street View. See section “Limitations” overleaf.
Mid-Term PlansWe are constantly working on improving this service. Therefore we expect regular improvements in the completeness and sophistication of the data we are able to deliver.
Pole MaterialBy Q3 2016, we expect to be able to deliver metadata describing whether a utility pole is made out of wood, metal, or concrete.
HeightBy end of year 2016, we expect to be able to deliver an approximation of the height of the pole from the surface of the ground. We may further estimate the pole’s true height (including depth below ground) to allow assessment of likely pole class.
Conductor AltitudesBy end of year 2016, we expect to be able to deliver an approximation of the height of each of the powerline and communication conductors attached to a pole. When a bundle of wires is attached, we will not attempt to locate the height of each individual wire, but rather the approximate average height of the bundle from the surface of the ground. We may explore additionally providing lateral distance from the pole surface - e.g. if a conductor is on a crossbar. We may explore heuristically estimating the type of conductor as electric or communications.
Number of Comms AttachersAs a feature requested by our Mexican customers, we may attempt to approximate the number of distinct communications parties attached to an electric-company owned pole as being zero, one, two, three, or more than three (as only three entities are allowed to attach to a pole). We don’t yet have an estimated date for when we can confidently deliver this data.
Has TransformerBy end of year 2016 we expect to be able to deliver metadata as to whether a utility pole has at least one electric transformer attached to it.
Has StreetlightBy early 2017, we expect to be able to deliver metadata as to whether a utility pole has at least one street light attached to it. We will not have a database of streetlights that are not also utility poles.
Strands & DropsWe currently have available estimates on which poles have strands that connect to which other poles, and what path given drops are likely to take either directly from a pole, or mid-span, but this is tuned to US-specific heuristics. We could discuss expanding this to other markets. We have found this useful for greenfield planning and auto-design in the United States.
APIIn the future, we may make available an API to allow for automated requests for poles in a given area, and for programmatic updates of progress in curating and delivering those poles, as well as pole data delivery.
Long Term Plans
Imagery AcquisitionWe are closely coordinating with Google’s Street View team on their next-generation imagery & LiDAR capture platform that should be able to provide us substantially more information about each pole, along with dramatically sharper imagery. We can provide imagery samples on request. This may allow features such as reading wire tags and street light wattage indicators that are not currently possible.
RemedyFor geographies requiring minimum conductor spacings (such as the US and Canada), we are investigating automatically suggesting whether a pole as-is has room to safely attach a new conductor, or whether a “wire plan” involving comms and/or power moves will be needed. This may be available in 2017 for select geographies.
MREWe would like to be able to provide full Make-Ready Engineering (MRE) survey quality information, including a 3D model of the pole and conductors suitable for importing into O-Calc Pro and/or SPIDAcalc. It will probably not be sooner than 2018 for us to provide this service.
PLAFinally, we’d like to be able to provide an engineering stamp-ready automated pole loading analysis of a pole, both as currently stands and with an additional conductor attached, suitable for submission to a permitting office. It will probably not be sooner than 2019 for us to provide this service.
Payments, Contracting, and Licensing
We will be happy to negotiate an appropriate agreement and data license so that you can use this product. Pricing may be either per pole, per linear kilometer, or per square kilometer. At certain price points we may be able to justify increasing the frequency or completeness of Street View and/or aerial coverage.
Limitations
Street View AvailabilityBecause the product uses Street View as its main source of information about utility poles, if Street View imagery is not available in a given area, we cannot provide pole data for that area. This would include private roads, gated neighborhoods, or areas Street View has not been able to access for other reasons. Customers can discover these areas on maps.google.com by dragging and dropping the yellow “pegman” onto a road to verify availability of Street View information. In Street View mode the date the image was taken is displayed in the upper-left corner of the screen, which will give an indication of the freshness of data available. If a pulldown to a slider is shown, it means imagery from multiple dates is available, which should give a sense of redrive frequency.
Rear EasementsUtility poles in “rear easements” - those located not along the road but rather in backyards - are not easily spotted from Street View and consequently depend on aerial imagery in order to locate the poles. Note that aerial imagery will not be as high quality as Street View imagery, and consequently we will be more limited in the level of data we can provide about such poles.
Accuracy Dependency on Aerial ImageryThe product uses aerial imagery to fine-tune position accuracy of located poles; in areas where there is higher quality aerial imagery available, we will be able to provide better location accuracy. If there is sufficient customer demand in an area for high position quality utility pole data, we may be able to make plans to acquire aerial imagery, and are happy to work with you on this. Similarly, we are happy to discuss increasing Street View coverage and redrive frequency for a given area if there is economic justification.
Satellite ImageryWhile Google does have access to satellite imagery, this imagery is not sufficiently sharp for us to be able to find utility poles.
Freshness / FrequencyWe are not able to ensure that data is fresh since we often rely on imagery that is collected not more than twice a year. So, for instance, if you needed a service that could let you know the moment a utility pole falls, our service would not be the right tool to use.
Transmission LinesOur product is not currently designed to index high voltage electric transmission lines. These poles often do not run along streets. If this is interesting to a customer, we are willing to discuss creating a product that includes such transmission towers, but please note that it is not currently included.
Altitude of BaseWe do not currently provide the approximate altitude above sea level of the base of the utility pole, but could discuss this with customers as-needed.
We’ve performed a full autodetection across all Street View imagery available for Canada, resulting in over 5m poles.
TeamIn addition to having many Google Fiber engineers working on this service, we have several people who are full time devoted to helping customers like you succeed with the Google Utility Poles Service. We’re happy to discuss the service with you by phone, videoconference, or in person at your offices. You should certainly also feel free to arrange a visit with us in lovely Mountain View, California for a demo and to discuss your needs. We’re looking forward to meeting you.
Please reach out to our team at [email protected] with your interest and needs and we’ll be in touch with you shortly.
David Weekly Product [email protected]
Kendra CommanderPartnerships [email protected]
Dave HubbardCustomer Delivery Software [email protected]
Bob PurvySenior Software [email protected]
Please reach out to our team at [email protected] with your interest and needs and we’ll be
in touch with you shortly.