First Google Maps and Google Earth let us spy on the world from the privacy of our computer screens. Now, a start-up is hoping to take mapping to a new level -- the street level, to be exact. Earthmine, based in Berkeley, uses technology developed for the Mars Rover to capture stereo panoramic images of streetscapes like this one in San Francisco and make them machine readable. Every pixel is associated with longitude, latitude, and elevation, allowing users to measure, say, the height of a building or the distance between your office and the nearest Peet’s Coffee. Founded in 2006 by John Ristevski and Anthony Fassero, the startup has raised $2 million and is working with alpha partners to refine its system. They envision a B2B subscription service with customers like construction firms, architects, emergency services, and partners that will develop applications for consumers. In this Q&A, co-CEO Ristevski, 31, talks about why the mapping industry is taking a quantum leap forward, how his system is different from Google maps, and why Earthmine cameras and low hanging cables aren’t a good mix.
Where did the idea for Earthmine come from?
Anthony and I were working at a nonprofit called CyArk here in the Bay Area. They specialize in high definition documentation of archeological sites and historical sites. They collect data using 3D laser scanners and created an archive of that online. We thought, Why can’t we do the same thing for cities?
There are just so many stakeholders that are interested in assets at the street level. There’s city, county, state government, homeland security, utility companies, people who maintain the road networks, insurance companies. They’re already maintaining very large databases of spatial information, but the information they have is literally dots on a map. The city of San Francisco has hundreds of thousands of signposts around the city but, believe it or not, they don’t have an accurate inventory. We can actually capture every single detail at the street level and make that available as a service.
To what degree will your database be searchable?
That’s really the key. How do we add intelligence to the database so people can find what they want quickly? We have a variety of strategies to help people get there. We’re using third party data sources like address databases and existing business listings to let you find that place. We're giving people some very easy tagging tools to let them tag things in 3D and we’ll make that searchable. Finally, we’re also investigating automated techniques that will allow us to find repetitive features in the urban environment that can be mass tagged and fully searchable.
How is Earthmine different from Google Maps or Google Earth?
We have a 3D infrastructure that supports these images. What really makes it powerful is you can click on the image and actually get a 3D coordinate -- latitude, longitude and elevation. You can measure, model, and tag things. With the Google system, it’s more a georeference that gives you a look at a place.
Do you see Google as competition?
At the end of the day, I think it’s more complementary. People will take all those different components and mash them up and create interesting applications.
For centuries, a map has been a piece of one dimensional paper. Are you, Google, and Microsoft reinventing the whole idea of map?
Absolutely. It’s one of the most exciting times in the geospatial industry in many decades, or hundreds of years. I think we’ll continue to see a revolution in the way people understand and interpret the world through these different types of maps.
Tell me about the technology you licensed from the Jet Propulsion Laboratory at Caltech.
We were looking for a solution to gather that very dense 3D data for a full spherical image. What the Jet Propulsion Lab had done on the Mars Rover is a very similar problem. On the bottom of the Mars Rover are two very wide angle cameras to map the terrain to help the Rover navigate the environment autonomously. We came across that research, and it was a very good fit for the problem we were trying to solve, which was how to collect dense 3D data that’s fully panoramic and spherical.
So technology developed for Mars is now exploring other strange planets -- like San Francisco and Berkeley?
Exactly. Those technologies can be directly applied to the mapping world. We’re not using that technology to drive our SUVs around San Francisco autonomously, but that data is being fed into our database which we can make available to people.
Describe how you gather these images.
We route the vehicle through every single street that’s accessible by the vehicle -- literally down every alley. It sounds like it would be incredibly slow, but because we can collect at regular driving speed, it’s actually reasonably fast. For one vehicle to collect the whole of San Francisco, every street and every alley, is about two weeks of driving. We think of the Earthmine vehicles as web crawlers for the real world. They’re going around and capturing the analog environment and turning it into digital database.
What’s the funniest thing you’ve found when you looked at the images?
I can tell you one funny story, which is very particular to this type of work. You can imagine people sharing illegal cable setups where they hang lines across the road. The Earthmine camera sits fairly high and when people hang lines across the road that sit a little too low, interesting things can happen.
You took out someone’s pirate TV cable?
Comcast should be paying us a commission for clearing all the illegal cable setups throughout San Francisco.
So far you’ve done San Francisco and a smattering of West Coast cities. How big an area would you like to do?
Our goal is to get to 50 major metropolitan areas in the next 18 to 24 months. We think that’s when it becomes interesting as a consumer platform.
What sort of commercial applications do you foresee?
If you have a 3D model of a building that you’re proposing, you could actually insert that into our scene and visualize it against the existing urban fabric. It’s something the city could evaluate to understand how your design fits in. In terms of construction, there are a number of ways you could use this information, mostly for reconnaissance. You could measure heights of buildings next to the lot, understand the types of buildings around the scene, understand the scale, measure the width of the lot, measure the offset from the street front, even the grade—basically everything until you get a survey crew on site. That’s something you can’t really do from a photo.
How about consumer applications?
The Earthmine system is really lightweight in terms of bandwidth and can be delivered easily over a mobile connection. You have your car navigation system that helps you get from A to B, but what happens when you actually get there? We think that the pedestrian navigation problem is a really exciting area that no one has solved. We’re seeing a lot of interest from potential partners and potential customers for that type of application. With the types of capabilities you have on cell phones, including GPS, accelerometers and digital compasses, the devices themselves are really primed to help you solve that pedestrian navigation problem. They just need that base layer of information to support these services.
Instead of giving me directions to a restaurant, you can literally show me how to get there by sending images to my mobile device?
I think you’ll see applications like that sooner rather than later. It’s just around the corner.