Inside AR: how augmented reality works

David Braue
30 September 2011, 6:00 AM


Your world, your way. We have a look at the hottest take on reality since virtual... only this time, it's personal.


It may have started out as a parlour trick, but augmented reality (AR) has rapidly become a global phenomenon by providing an impressive, flexible new way to view the world. By overlaying physical views of the world around you with up-to-date information, everybody from municipal authorities to real estate agents and artists are finding new ways to engage with customers and audiences.

The first time you see AR in action, it's almost magical: from the ball-juggling action of ARSoccer to imaGinyze's lane and speed-tracking Augmented Driving to the plane-finding capabilities of pinkfroot limiter's Plane Finder AR, developers have found all sorts of ways to make intelligent overlays that give you more information about your surroundings.


Augmented Driving watches the road in front of you, marking lane boundaries and alerting you when you are speeding or coming close to nearby cars.

Indeed, Apple's iTunes App Store features over 180 iPhone AR apps and the Google Android Market shows over 500 such apps, covering everything from video games to subway finders, novelty applications to intensely useful tools. Restaurant guide Urbanspoon has added AR features to its app, while the CommBank Property Guide tells you the purchase history and vital statistics of any house at which you point your phone.

Many are purpose-built and serve a specific purpose, while others function as layers on top of AR browsers like the Layar Reality Browser, Wikitude World Browser, WhereMark, Yelp, junaio, and others. By providing an open platform for the creation of augmented-reality layers, these apps have democratised the AR industry - and made possible flights of fancy such as Unseen Sculptures, in which new-media artist Warren Armstrong and nearly two dozen others collaborated to populate central Melbourne and Sydney with a range of virtual artworks that "appeared" when smartphones running the Layar app were pointed at particular parts of the city.

"I'm fascinated by mobile devices and their ubiquity," says Armstrong, who has spearheaded the project's expansion - including planned engagements with the Sydney Fringe Festival, Cairns Festival and other projects. "It's very immersive: you can go to a location, put on headphones and do amazing things with sounds. Layar takes care of the fiddly back-end stuff for people who aren't au fait with PHP or MySQL programming. The whole thing is in its infancy, and there's a lot of scope for creating new things."

Build your own reality

AR has been around for decades, but the rise of smartphones - which bundle necessary sensors and software capabilities into a mobile platform - has given it critical mass. At its heart, all AR engines work in largely the same way. Using GPS, compass and gyroscope readings, they continually monitor your location, direction, and orientation in the real world. By comparing your location and directional position with a pre-built database of points of interest (PoI), they can calculate the distance to those PoIs and overlay them onto the image fed to the app by your smartphone's camera.

The result - whether unseen artworks or a sci-fi heads-up display (HUD) - is akin to walking through an interactive version of Google Street View, fed by PoIs from Google Maps' various layers. Indeed, there is considerable overlap between Google's ubiquitous mapping applications and tools used to make AR worlds: users design augmented-reality landscapes using 3D design tools like Google SketchUp, which Google designed for bundling 3D objects into Google Earth's kml files. Export your SketchUp objects into the Wavefront OBJ format, and you can load them straight into an AR environment for use within a browser such as Layar's.

Layar calls on other standard data formats to represent objects in users' AR creations: for example, including JSON (JavaScript Object Notation), a standard data-interchange format that organises pairs into name/value pairings or lists of values. JSON allows the location of individual PoIs in the augmented-reality world to be stored in a standard format for easy access and comparison to the user's current location. When a user's application issues the Layar API with a GetPOI request, the service uses JSON to return a list of nearby PoIs that is rendered by the client application.


Point the Commowealth Bank Property Guide app at a house, and it uses AR techniques to pull up its vital stats and sale history.

With 1,000 active developers having already created around 2,600 layers that are available to users, Dirk Groten, chief technology officer with Amsterdam-based Layar, says the platform is rapidly growing in popularity - especially since it was chosen by Samsung to come pre-loaded on the company's popular Galaxy S II smartphones; as a result, 75% of Layar's 1.5 million active users are running the Android version of the app.

Although the AR browser app provides users with their interface into the virtual worlds you've made, content is hosted on the content provider's own servers; the browser is an aggregator of third-party content links, just as iTunes points to third-party podcasts but leaves that content on the third-party servers. Layar does cache a set of PoI icons for each user - but the actual live data is stored at the user's end, whether this means a hosted corporate server or just a file shared from an individual's Dropbox account.

Layar's position as middleman in the transaction not only allows it to publish a library of available AR layers, but allows it to ensure links don't point to non-working URLs. "We've never rejected anything for publication on the basis that the content wasn't good," says Groten, "but we have rejected numerous things on the basis that it just doesn't work."

The catalogue approach was chosen for more than just convenience on the browser maker's part: it allows developers of AR layers to build in a broad range of interactive features - for example, live Twitter feeds, updated PoI databases, and new functionality or in-world objects.

"The best Layars are the ones that are dynamic and that are actually more like web apps," says Groten. "It's very difficult [for us] to make something scalable that would be dynamic. This way, developers can script in whatever language they want and take responsibility for access to Twitter and other APIs."

AR for the future

Although early AR applications were largely novelties, the concept is catching on both in artistic circles and in the commercial world. A number of AR browser makers are rejigging their platforms with a set of APIs and a software development kit (SDK) that allows AR worlds to be embedded inside other commercial applications - removing the need for the user to run a specific browser and manually locate the AR layer in question.

Other apps are looking for new ways of identifying users' locations: junaio, for example, offers image recognition that lets it not only recognise specific objects, but 'glue' a 3D model onto an object seen through your smartphone's camera. The 3D model stays in the correct location and orientation even as you move the smartphone's field of view. This would be hugely useful, for example, in a tourist application highlighting points of interest or an historical application using graphical overlays to show what a neighbourhood or building looked like at different points in the past.

Another AR browser, Wikitude, has taken a social approach by allowing registered users to drop PoIs onto a collectively-maintained map; each PoI can be linked with URLs and other information, then viewed by users of the Wikitude World Browser. Wikitude has already built on this platform to offer Wikitude AR Travel Guide, an Android app that picks out nearby points of interest for travellers.

A range of other AR toolkits are also available. The open-source ARToolKit, for example, runs across Windows, Mac OS X, Linux, iOS, Android, Symbian and SGI IRIX operating systems, and uses image analysis to pick out key points of interest and overlay augmented-reality objects; FLARToolkit, a spinoff, is a port to Adobe Flash that works in a Flash-enabled web browser. BuildAR offers free and commercial versions enabling creation of 3D models and scenes; other toolkits include ArUco (Windows, Linux, BSD), mixare (iOS, Android), OpenMAR (Symbian), Argon (iOS), Goblin (Microsoft XNA), and PTAM.

The key to AR's success, however, will be the creativity of its users: "We think AR can become very big," says Groten, "but it cannot become very big on its own. All the experiences you see are part of a campaign [by creatives]. It's always part of a bigger thing. AR is not the end; it's a means."



Post your comment



anonymous user Anonymous user