Earthmine: Building a 3D Datamine of the Urban Environment
Earthmine, the Best Technology Innovation/Achievement category winner at tonight's Crunchies, is a company that might seem uninteresting at first glance. When I first saw earthmine I assumed that it was just a Google Maps Streetview knock-off. I was wrong.
This startup is doing something far more interesting than that. While Google Maps and related consumer products have whetted the public's appetite for visualization of specific places on a map, earthmine is making those places machine readable.
How it works
The company uses a proprietary array of still-images cameras to take photos in stereo at regular spacial intervals while driving through city streets. The resulting 3D images can be measured with an accuracy that corresponds to measurements of the physical objects and distances they represent.
The company says it covered San Francisco in just three weeks. Each day's data is processed automatically and is available before the next day begins.
The initially self-funded company recently took an investment from CalTech and secured an exclusive liscence to use 3D image processing technology developed by NASA's Jet Propulsion Laboratory (JPL). Generating dense, accurate 3D data from wide-angle images is a serious technology challenge but one that the JPL worked on to process data returned by the Mars Rover.
What it means
Just as we here at ReadWriteWeb are excited about the potential offered by a machine-readable, or semantic, web - so too are the possibilities countless when thinking about a data rich, accurate and machine-readable 3D representation of the urban environment. earthmine offers a usable looking web interface but that's just the friendly wrapper around a dataset of far greater consequence.
From urban planning to mobile services to security applications, this kind of data and interface has a lot of potential. If the value of mapping and of GIS are clear, the value of a geospatial 3D dataset about urban environments should be clear as well. Combine all three and you'll be able to assemble some very interesting resources on almost any topic.
It is important to me to say that I don't care for the way the company talks about the technology, as "reality mining" and "indexing reality." To call that tasteless would be an understatement. I'm concerned that such reductionism could have substantial adverse political consequences. Maybe I'm just old fashioned to believe that there's far more that's important in "reality" than the things that can be digitized - and that much of it ought not be mined. I should probably stop, though, before a corporate exit puts me in thumbscrews listening to a well-fed Dr. Evil laugh. This technology itself could be put to use for good or ill, I'm sure.
Either way, this is fascinating stuff and worth some thought no matter how you relate to it. In addition to the very well produced company-produced video below this interview with the young earthmine CEO and this one of his time on stage at the DEMO Fall conference is worth a watch.
Comments
Subscribe to comments for this post OR Subscribe to comments for all ReadWriteWeb posts
Not sure I understand why this is Earth-shaking - it must be the measurement technology that makes the technology special, or is that spacial?
I also don't understand the novelty of this. How is it different to what Google have been doing?
For one, the Immersive Media Corp camera (the company used by Google for the "street level views") only records in a max resolution of 640x480 in motion. Then stills from the video feed are used - hence blurry and poor quality images.
Earthmine shoots "in stereo" - ie more than one camera, in motion, in a much higher resolution than Google Streetlevel.
The "stereo" mode allows precise measurements be made - the one example from the video shows a base measurement from curb to center lane, then shows the difference in distance with a vehicle too close to the center line. Apparently you can also determine (not only latitude and longitude) but elevation as well.
I don't see this tool being of much use other than curiosity for the general public, however it has amazing potential for architects, engineers, urban planners, municipalities, repair crews, in other words folks with a project oriented need to see this kind of data. What would REALLY be neat is to be able to virtually "knock down" existing buildings digitally or "draw in" non-existant buildings, to scale, to effectively maintain any "skyline bylaws" prior to any costly construction projects.
If one reviews the Immersive Media Corp's website you can see that the streetlevel views project began with a media release in 2005. The photos were collected since then. The photos we see today on Google's Streetlevel are potentially up to 2 years out of date. With a 3 week turn-a-round to photograph San Francisco, Earthmine looks like they are providing a much more recent view than the google product.
The panoramic photograph is as old as photography itself. The full motion panoramic video camera has been a project in development since at least 1998, just about a decade. Digital photography has made panoramic imaging more accessible to a wider audience, so really, there is absolutely nothing novel about the technology - its precisely the application of a panorama in relation to everything else that is the real issue here. Novelty is nice and all, but making tools is way cooler.
I suppose the spectre of "privacy" is likely going to be raised again - however this time the issue will be whether it is better to have 3 week old civic views made available only to the select engineers and contractors (or law enforcement agencies), or to the general public - is it safer only to let a few people spy on your city, or better to let everyone have the same chance to see it so as to avoid being considered a form of lukewarm espionage. Homeland security is touted as one of the potential applications on the Earthmine site.
Google are using Point Grey cameras.
Google has been using point grey cameras since 2008. 21 days.
http://www.immersivemedia.com/details.php?id=161
Immersive media began the process in 2005 though:
http://www.immersivemedia.com/details.php?id=31
Thanks for bringing up Immersive Media. I just wanted to take a minute to clarify IMC’s technology.
Immersive Media’s camera has 11 lenses (11 CCD’s) simultaneously recording at 30 frames per second. Each lens records at a resolution of 640x480, and the overall stitched resolution is 2400x1200. Each frame of video is embedded with highly accurate GPS coordinates and other metadata. Our submeter accurate GPS data allows us to provide a photogrammetry solution using only our single multi-sensor camera.
On image quality, because of the compression that takes place for online streaming, the flash video resolution is not as high as the final footage that we deliver to our customers (city planners, engineers, mapping applications). However, pausing the streaming flash videos will display a higher resolution jpeg. With the release of Flash 9, all of our videos on our demo page have been upgraded to the fully spherical flash viewer, as opposed to the panoramic viewer we were limited to previously in flash. http://demos.immersivemedia.com Our videos are also embeddable.
Immersive Media did begin collecting imagery in 2005 with about 100 miles in each major U.S. and some Canadian cities. Since then, we’ve gone back to all of those major cities and collected over 1000 miles in each, in addition to some smaller cities. As we all know, cities are constantly changing and new images need to be collected regularly. We’re currently recollecting some of the biggest U.S. cities and will continue to update them regularly. So, although we did start collecting cities in 2005, we’ve been constantly updating our database ever since. I assure you, the data we collect is much, much newer than three years old, as we’re adding new streets to our 50,000 + mile collection every day.
For any questions, don’t hesitate to contact us through our website.