Skip to Content

Indoor Imagery Shows Mobile Devices the Way

Street View-style imagery of interior spaces lets mobile devices locate themselves more accurately than is possible with GPS.
December 10, 2013

Smartphones locate themselves outdoors using a GPS sensor, but those signals are blocked indoors. A new technique uses a device’s camera to get an indoor location fix to an accuracy of within a meter. The technique could enable new kinds of apps, and may be particularly valuable for wearable computers such as Google Glass.

The new location-fixing method is being developed at the University California, Berkeley. It uses a photo from a device’s camera to work out the location and orientation of the device. It does this by matching the photo against a database of panoramic imagery of a building’s interior, similar to the outside views offered by Google’s Street View. The system can deduce the device’s location because it knows the position of every image in that database.

The researchers used a special backpack that captures Street View-style imagery indoors as the wearer carries it around. It has two fisheye cameras, laser scanners, and other sensors. Software uses the data collected to generate a map of the building’s interior, a stitched-together set of panoramas, and a database of individual images that can be used for location lookups.

“You can provide that blue dot you see on a mobile map when out-of-doors for interiors,” says Avideh Zakhor, who leads the Berkeley group developing the system. Zakhor previously sold a 3-D city mapping company to Google that became a major part of the company’s Google Earth 3-D virtual globe.

Zakhor and colleagues have tested their system in buildings on the Berkeley campus and in a mall in Fremont, California. In tests at the mall, they successfully matched more than 96 percent of images taken by a smartphone’s camera against the database of images, the researchers report in a paper on their tests. When the matches were turned into location fixes, most came out with an error of less than a meter from the device’s true location.

Zakhor says her approach compares favorably with competing methods of determining location indoors in terms of accuracy and the cost of deployment. Alternative methods include using Bluetooth “beacons” or fingerprinting the pattern of Wi-Fi signals inside a building.

Jonathan Ventura, senior researcher at Graz University of Technology, Austria, agrees. “The major advantage of image-based localization is that it works almost everywhere and doesn’t require changing the environment in any way,” he says.

Zakhor’s group isn’t the only one capturing such data: Google has begun taking its Street View product inside and announced last month that it had documented the interiors of 16 airports and over 50 train stations.

Ventura’s own research focuses on augmented reality. He says that if devices can be located very accurately it will allow for virtual and real worlds to be closely aligned. “If we want to render a rich and complex virtual world into a high-resolution image,” he says, “we need to have much more accurate positioning than a consumer GPS receiver can deliver.”

Zakhor is planning tests of her method on computerized glasses, with the intention of having the devices use snapshots to track their location, making it possible to provide a map of an interior space in a person’s field of vision. The Berkeley research group is also working on using data from Wi-Fi signals collected by their backpack to provide a secondary method of deducing a device’s indoor location.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.