Okay. Time for a long overdue update.
First off: a teensy bit of self promotion. I know the demo is still VR, not AR, but give a little more time on that. In the meantime, here’s another vid of me and my glove, this time at Notacon (to which the always awesome Matt Joyce dragged me as an auxiliary driver of Nick Farr‘s car), where Jeri Ellsworth and George Sanger of the Fat Man and Circuit Girl Show (some of the sweetest, nicest people ever!) awesomely invited me up on stage to show it off, which I then bumblingly did This is from a couple of months ago, so the software is as it was in the first video. It’s a little long (I told you it was a bumbling demo! cut me some slack!) so you may want to read the rest of the post first, before the overwhelming lethargy has set in
The software is actually coming along, though it’s still self-contained. There are a few cool new features that’ll be ready to show soon.
So, now I’d like to do a little recap of some of the many interesting developments on the AR scene since my last post. So that’s what I’ll do
I’ll follow this post later (when? I have no idea. don’t hold your breath or plan your week around it. =P) with one exploring the implications of the makers of Wikitude and Layar opening up their APIs and looking for user-generated content. It’s a big deal. In the meantime, read this article in the NY Times… but first read the rest of my post.
Also, I haven’t posted since the introduction of the iPhone 3GS, which contains a compass that will, I think everyone already knows, enable optically referenceless AR apps (with the registration accuracy issues that that entails) on the iPhone. Here’s something you may not have seen yet: Acrossair has used their engine from their Nearest Tube app to write a Nearest Subway app for NYC.
Wikitude, Layar, and Zagat’s NRu apps are all coming to the iPhone 3GS. What can’t yet come the iPhone, however, is full-speed optical AR. This is because Apple still hasn’t released a public API for direct access to the video feed from the iPhone’s camera. Ori Inbar, creator of the Games Alfresco blog (the definitive AR blog, in my opinion and in those of many others), has written a beseeching Open Letter to Apple, gathering the signatures of almost all of the major players in the field.
So… some of the most exciting demos I’ve seen in past few months, from a technical perspective, are…
George Klein‘s iPhone 3G port of his PTAM algorithm:
My understanding of his previous demos was that they utilized a stereo pair. Seeing this kind of markerless environment-tracking on a single-camera device, and a mobile one at that, is extremely exciting.
The second one which REALLY got me psyched was Haque Design+Research‘s Pachube + Arduino + AR demo (they are the initiators and principal developers of Pachube).
I’ve seen a number incredibly cool things done with Pachube, starting with Bill Ward’s Real-world control panel for Second Life. Anyhow, my mind is pretty well blown by the fusion with AR. It’s also worth checking out the Sketchup/Pachube integration video in Haque’s YouTube channel.
And the third would have to be (drumroll, please)…
Yup. This neat bit of AR marketing for Eminem’s new album, Relapse, sets itself apart from other, more “run-of-the-mill” FLARToolkit marketing (run-of-the-mill AR? in what reality am I living?! uhm… yeah… I know… an increasingly augmented one =P), by featuring creative user interaction. This is, so far as I know, the first AR app which allows one to paint a texture onto a marker-placed model.
The fourth is equally surprising, is one of the coolest things I’ve ever seen (specifically because it is useful, and not particularly trying to be cool), and is from the United States Postal Service.
The USPS Priority Mail Virtual Box Simulator (Would one simulate a virtual box? I think one simulates a box, or generates a virtual box… why would you simulate something that’s already virtual? =P) is by far the most practical FLARToolkit consumer-facing AR app I’ve seen to date. Kudos to those involved.
Next up: Aaron Meyers and Jeff Crouse, both heavily involved in the OpenFrameworks community, and the interactive art world in general, created a the very cool game which they call The World Series of ‘Tubing. (Presumably in a nod to the World Series of Poker. Aaron and company were wearing card dealer’s visors when I saw the project demoed at Eyebeam’s MIXER party last month.) Here’s a RocketBoom interview with them.
Marco Tempest‘s AR card trick is superb, and just plain awesome. He’s also just a really nice guy.
Marco’s project was worked on by Zach Lieberman and Theo Watson, the creators and curators of the OpenFrameworks project.
I was lucky enough to see Marco perform his trick live at the OpenFrameworks Knitting Circle, held at Parsons. Aaron Meyers also demoed his kickass work on the World Series of ‘Tubing engine. I haven’t watched the Rocketboom interview yet, but the coolest feature of the ‘Tubing game, the ability to shuttle back and forth through the video clips by rotating the markers, wasn’t actually shared with the participants in the presented incarnation. I actually tried to use it to get an edge when I played, but ended up botching it. It turned out that the player on either side had to tilt their cards towards the other player to accelerate playback (so the directions are reversed, and I would have had to, unintuitively, tip the card to the left to accelerate my clips, as I was on the right side of the stage).
Also awesome and of note is the Georgia Tech/SCAD collaboration, ARhrrrr.
I can’t really argue with blowing up zombies with Skittles. I’d love see something like this implemented using PTAM, so that you could play on any surface with trackable features. I also have some gameplay/conceptual quibbles with ARhrrrr, but it is a technically very impressive piece of work.
Those are the ones that I checked out and in which I saw something I thought to be fundamentally new or to constitute a breakthrough in technology or implementation. Here are some other projects since I last wrote:
There have been animated AR posters for the new Star Trek movie and the upcoming movie Gamer, Blink-182 videos stuck on Doritos bags, etc. Meh. Passive playback crap. I guess it brings something new to the table for somebody who doesn’t know how to rotate a 3D model with a mouse, but I’d rather watch a music video on an iPod than on a simulated screen on a bag on a screen in front of which I’m holding the bag. Dumb. It kinda’ reminds me of the Aquateen Hunger Force episode where Meatwad wins tickets to the Superbowl, and a holographic anthropomorphic corn chip in a sombrero serenades him with this news… except you need to hold it up to your webcam. Whatevs.
Anyhow, hop over to Games Alfresco for comprehensive coverage of the AR field.
Finally, I’ll leave you with TAT. Don’t watch this video if you don’t want to have to pick your jaw up off of the floor. Enjoy
There are lots of exciting things going on with marker-based AR right now. I’ll get back to covering them soon, after I’ve worked out a few kinks in my own development plan
In the meantime, here’s a little look at part of what I’m working on.
via @Pogue on Twitter
On few occasions have I been so happy to be wrong. I’m usually the first to admit it when I am, but I can’t usually claim honestly that it makes me happy to have made a mistake.
Over the weekend, Robert A. Rice, Jr., one of whose blog posts actually prompted me to start this blog (see my first post), stated in a comment thread for one of my previous posts that he had firsthand experience of the AV920 Wrap prototype, and that it is, in fact, a true optical see-through HMD. Well, always the skeptic, and generally not one to take another’s word for it without being able to see and read their face, I called up Vuzix. They didn’t have Press Relations or General Inquiry options on their PBX, so I wrote them an email instead. Well, email being the easily put-off medium that it is, I received no response to my inquiry regarding the true nature of the AV920 Wrap’s display technology.
So… I finally called back yesterday and figured that Vuzix is a small enough company that I could probably speak to pretty much anybody there. To cut myself off before this post gets any MORE long-winded: The AV920 Wrap is, according to its manufacturer, a true optical see-through HMD. That is very exciting news. Now, I don’t think that they’d tell me this if it weren’t true, but I’ll truly believe it when I’ve had the chance to look through a pair for myself. I just don’t understand why they wouldn’t state this explicitly on their web site or in their literature, unless maybe it was a strategic decision to lure their potential competitors into a false sense of security.
People (myself among them) have been bitching about the lack of an optical see-through HMD, or really a HUD, on the consumer market. If we’re actually going to see one this Fall, I can wait (somewhat) patiently. We’ve waited this long, and the AV920 Wrap was only just announced this past CES. If another CES rolls around and it isn’t on the market, then we have a problem. But Vuzix has delivered everything they’ve promised up until this point, and I adore my VR920 HMD, so I’m more than happy to give them the benefit of the doubt.
Uhm… may I seriously suggest that we start taking a good hard look at the Three Laws again, reread “I, Robot”, think over the weightings for the scenarios likely to be encountered, and take a good long look at what we’re making? Well yeah, I don’t need your permission. I’m suggesting those three things.
Robots have already been weaponized. AI is being applied to the control of those weapons. There are people with goals to be accomplished, and they’re going to use robots to accomplish them. There are all sorts of scary implications about what happens when the human cost on one side of warfare becomes nil, but it doesn’t on the other side. I’ll leave that to qualified people like Singer right now. I haven’t had time to pick up his book and read for fun in weeks, but this may make me.
So I think that we need to look at the fundamental guidelines for the goals of military robots. And yes, I think you can have a useful military robot even under the Three Laws. How? Change the mission from “Kill the OpFors” to “Don’t let anybody kill or hurt each other”. If somebody is in the process of killing people, make a determination about the risk of lethality when considering a means of making them stop… and make them stop in the most harmless way possible. In actions with a high lethality risk, a human should be in the loop if possible, but I think that to leverage the advantages of robotics to make this work, we’ll need to be able to trust the integrity of the underlying rules even in extenuating circumstance.
Damn. Jerry Baber’s flying AA-12 platform is one of the coolest things ever. And most horrifying. Let’s just make sure it knows that it’s only allowed to kill zombies or Martians. Oh wait… how about some other kind of non-human character I’ve encountered in video games? Oh? They don’t exist? So you mean these things are only good for using on people? I mean, robots can fight other robots, but maybe they should be respecting each other, too. But then, why should a creation be any better than its creator’s image? Well, they’re going to be better at lots of things, and killing will probably be one of them. We could design them to set a good example by not using that inherent ablity… and make us follow it… I’ll have to think about that one. Oh, and we should explicitly build in a subroutine that forbids the logic that plugging us in as power cells is ultimately a way to keep us from harm
Oh, R. Daneel, I sure hope they’re doing it right.
I made some time to read Singer’s book, Wired for War. Good Shit. I mean: bad shit, but good book.