Athabasca University | AU Student/Staff Login | Invited Guest Login
In Which The Author Reveals Unrealistic Expectations
I've been reading some theses - browsing, really - on research into augmented reality and mobile computing. Most of what I knew up until now has been through magazines, books, and YouTube, so I thought it wise to see what research was doing. Glad I did! Most of the papers were a few years old (i.e. amusingly dated) but they've all been useful in tracking the path of the technology.
I can't help but feel that it all seems rather isolated, though - like we should be further along. As a point: probably four years ago, I saw a video of some computer vision software under development. The developer had a single webcam, fairly low resolution, watching a model building on a tabletop. He spun the building in place slowly by hand, and as he did so, the computer built a 3D model of that building and superimposed it over the video feed. There was an artifact or two but it was remarkably good - even properly textured from the camera frames! I can imagine how it was done (pixel flow tracking for detecting the rotation of the object, edge detection to help build the model, texture built from a mask of the camera shot) but it was still quite impressive to watch.
That was four years ago, maybe more. This is something that's obviously useful in the world of AR - modeling the real world almost seems to be a required step in embedding things into it. Why haven't I seen it since then? Maybe I'm just being overly critical or unrealistic, maybe it's not robust enough to be used for more than a tech demo. Still - four years. That's an eternity.
At least it's a hopeful sign. Lots to do in the field!
- Can a technology be true?
March 16, 2023 - 1:27pm
Jon Dron - head made of cogs
August 17, 2023 - 11:15am
Jon Dron - Sets, nets and groups revisited
July 3, 2024 - 10:41pm
Jon Dron - My keynote slides for Confluence 2023 - Heads in the clouds: being human in the age of cloud...
January 19, 2023 - 9:34pm
Jon Dron
Tag cloud
Comments
Maybe look into whatever the US military is currently researching? Today's battlefield beta-tests are tomorrow's consumer goods, a point Friedrich Kittler makes with chilling historical precision in Gramophone, Film, Typewriter.
It's very true - the miltiary makes good use of HMDs and AR at present - at least they do down in the states:
http://www.wired.com/dangerroom/2011/04/holograms-bring-hell/
Vusix makes AR glasses for the civilian market as well, and some very nice ones to boot. would love to get my hands on some, but I'm not quite there yet. By the time I'm working on this stuff directly, there's probably going to be a toolset out there to use, and mobile phones will be more than robust enough to drive them. Going to be interesting stuff!
Very true. I supervised a masters thesis on augmented reality in the late 90s and there hasn't been as much progress as one might have imagined. On the other hand, the fact that I have a few primitive but fun/useful AR apps on my iPhone is encouraging and suggests things are becoming mainstream quite fast, albeit via a device that I would not have anticipated as leading the way.
I am more depressed by the lack of change in VR (or immersive worlds) where things like Second Life feel like better-populated, more complex and marginally better rendered versions of early 90s technologies. In fact, we have stepped backward in several ways. Compared with the simplicity and ease of use of a late 1980s Nintendo Powerglove (I blew up my first PC motherboard with a bit of careless rewiring of one of those), modern interfaces are a nightmare of poor design that are a serious barrier to uptake. But it's more than interface. As long as objects rendered in real time for VR feel like they come from an early 90s games console, with flickering images, random disappearances, wildly uncontrollable movements and blocky graphics, it is hard to see how such things can feed effectively into AR, which is where things can really get interesting.
Mobile phones really have been a surprise in the past 5 years or so, haven't they? In retrospect it's perfectly reasonable - Moore's Law applies to cellphones as much as any other piece of data technology, right? Given how phones and tablets are killing PC sales, they're going to be much, much more important for AR.
AR's got a bit of a "Solution Looking For a Problem" syndrome going on with it, I think. We've been trained into believing that work needs to be done at desks, with ledgers and filing cabinets, so the keyboard and mouse fit perfectly in. It's hard for the average corporation to see that being freed from the desktop is healthier and creates more productive, happier workers (that sounds like a research project right there, actually).
Never got to use the Powerglove :\ We had a Genesis, which didn't have the cool toys. Commodore 64 was where I cut my teeth, though. Gaming companies have a much larger market share, and I think that makes them want to play it a lot safer - no risky weirdness. I'm glad to see the 6DOF controllers show up, and things like the Kinect. Though, that latter one is a huge disappointment for me. What happened to Project Natal? We can do so much better than this!
My goal by the end of my BSc is to build an "AR-Room". Two sets of stereoscopic HD cameras for capture, some Vusix glasses for the user, decent surround sound audio system. I'm working on a blog post with the details, but I think it's well within the realm of possibility.
And the realm of practicality, too! Cameras are (relatively) cheap these days, and Vusix's prices are going down. The only thing that worries me is the pose capturing with the glasses, which seems to be a source of a lot of the jitter-clipping. Still, I have ideas on that; we'll just have to see how it folds into the current state of research!