Google Again
AR or Augmented Reality needs one very important piece of information if it is to provide a service to consumers, and that is where it is. While you can see where you are through the AR lenses, if the system is to project other images or data onto your actual image, it has to know where in space the glasses are to position things correctly, especially directional information, which we expect to be an integral part of AR for consumers. Without location data the AR system will not be able to point you toward a particular destination or help you pick out your rideshare in an airport waiting line, and more importantly, it could point you in the wrong direction if being used in a driving situation. At a recent conference, Google (GOOG) highlighted its work in what it calls Google Live View for developers, which is based on Google’s Street View database, a 20m+ GB collection that comprises satellite imagery, geological surveys, municipal maps, and the 10m+ miles of roadway taken by its Street View cars, all combined to produce accurate maps, turn-by-turn navigation, business locations and real-time traffic information, much of which is sourced from GPS enabled smartphone users, all of which will be licensed (API) to developers who wish to use this incredibly accurate mapping data.
While we expect Google will certainly be a player in the AR application market, the company is limited in what applications they can develop, both by resources and by the necessity not to undercut their client base, so by licensing this vast data resource, the API gives developers direct access without having to build their own interface which might include built-in sensors or ‘anchors’ that would have to be pre-defined and expensive and completed to develop on their own. Google seems to have stepped forward as the de facto supplier of the geospatial information needed to make AR work, and is already using the API as part of its Live View mode inside the Google Maps application and as the AR Places filter in Google Lens. The idea of the API is to free developers from having to build out their own spatial information interfaces by just using (and licensing) the Geospatial API.
GPS is commonly used for location data, but the accuracy of that data is not always accurate, especially in densely built-out areas, leading to between 5 and 10 meters of positional accuracy and 35 to 40⁰ of rotational accuracy, which could lead to AR images being out of view or the necessity for consumers to place anchors at a particular location to allow the system to have a point of reference, which might work well in a small space but not in a larger venue By using the Google Geospatial API all of the amassed location and image data that Google has collected over the last 15 years becomes available to the developers application, at least anywhere where Google has StreetView data (not in Germany, North Korea or China and a number of other countries), which gives almost pinpoint accuracy and an easy way to guide users to location where they might spend some money.
After spending some time understanding how the Google API works from a technical standpoint, we were even more impressed than just listening to the company’s promo. Without going into detail, the API links the user’s mobile camera to the company’s servers and matches the cameras image to the Google data. Given that the StreetView data has been filtered by Google AI to eliminate non-stationary objects like cars and people, the algorithm matches the camera image by comparing against the buildings in its database, which includes buildings all over the world. Once a match is made, the detailed coordinates are sent to the API from the Google servers and the AR application can then accurately overlay the new information on the camera image. Considering the billions of buildings the algorithm has to ‘look at’ to match the camera image, the technology is quite incredible
While it might seem that we are favorable toward Google, as we have singled out their AR translation application previously and the API mentioned above, Google has amassed such a large volume of disparate data that it seems able to coordinate, that they can easily create applications and data for developers and customers that is unavailable from other sources. In reality our favorable view of Google from an AR point of view comes from the fact that the company not only collects user information, as many social platforms do, but has been collecting other, sometimes seemingly strange data, for so many years that its databases are the most fertile grounds for AI learning, whether it be image related or data related, so the edge that it gives the company for things like the positional data mentioned above or the translation data we have mentioned in the past is a distinct advantage over other data collectors that have not been doing so for an extended period or to the extent Google has.
As the globes primary search engine, the company has had access to so much consumer data that would likely be unavailable to others, it puts the company in the sights of those concerned with anti-competitive behavior. But the fact that the company has had the foresight to spend the time and resources to collect what might have seemed irrelevant data is more the reason why it was collected and not as much for the world domination that some consider Google’s goal. They had the opportunity and were willing to allocate the resources likely without the ultimate goal of using it for such specific applications. That’s smart thinking, and while it might lead to world domination in data resources, it was good planning years ago that made it possible.
While we expect Google will certainly be a player in the AR application market, the company is limited in what applications they can develop, both by resources and by the necessity not to undercut their client base, so by licensing this vast data resource, the API gives developers direct access without having to build their own interface which might include built-in sensors or ‘anchors’ that would have to be pre-defined and expensive and completed to develop on their own. Google seems to have stepped forward as the de facto supplier of the geospatial information needed to make AR work, and is already using the API as part of its Live View mode inside the Google Maps application and as the AR Places filter in Google Lens. The idea of the API is to free developers from having to build out their own spatial information interfaces by just using (and licensing) the Geospatial API.
GPS is commonly used for location data, but the accuracy of that data is not always accurate, especially in densely built-out areas, leading to between 5 and 10 meters of positional accuracy and 35 to 40⁰ of rotational accuracy, which could lead to AR images being out of view or the necessity for consumers to place anchors at a particular location to allow the system to have a point of reference, which might work well in a small space but not in a larger venue By using the Google Geospatial API all of the amassed location and image data that Google has collected over the last 15 years becomes available to the developers application, at least anywhere where Google has StreetView data (not in Germany, North Korea or China and a number of other countries), which gives almost pinpoint accuracy and an easy way to guide users to location where they might spend some money.
After spending some time understanding how the Google API works from a technical standpoint, we were even more impressed than just listening to the company’s promo. Without going into detail, the API links the user’s mobile camera to the company’s servers and matches the cameras image to the Google data. Given that the StreetView data has been filtered by Google AI to eliminate non-stationary objects like cars and people, the algorithm matches the camera image by comparing against the buildings in its database, which includes buildings all over the world. Once a match is made, the detailed coordinates are sent to the API from the Google servers and the AR application can then accurately overlay the new information on the camera image. Considering the billions of buildings the algorithm has to ‘look at’ to match the camera image, the technology is quite incredible
While it might seem that we are favorable toward Google, as we have singled out their AR translation application previously and the API mentioned above, Google has amassed such a large volume of disparate data that it seems able to coordinate, that they can easily create applications and data for developers and customers that is unavailable from other sources. In reality our favorable view of Google from an AR point of view comes from the fact that the company not only collects user information, as many social platforms do, but has been collecting other, sometimes seemingly strange data, for so many years that its databases are the most fertile grounds for AI learning, whether it be image related or data related, so the edge that it gives the company for things like the positional data mentioned above or the translation data we have mentioned in the past is a distinct advantage over other data collectors that have not been doing so for an extended period or to the extent Google has.
As the globes primary search engine, the company has had access to so much consumer data that would likely be unavailable to others, it puts the company in the sights of those concerned with anti-competitive behavior. But the fact that the company has had the foresight to spend the time and resources to collect what might have seemed irrelevant data is more the reason why it was collected and not as much for the world domination that some consider Google’s goal. They had the opportunity and were willing to allocate the resources likely without the ultimate goal of using it for such specific applications. That’s smart thinking, and while it might lead to world domination in data resources, it was good planning years ago that made it possible.
RSS Feed