Atlas Kickstarter

[djesko]

http://www.kickstarter.com/projects/623 ... -made-real

Strap your smartphone to your chest and use floor markers to wonder around your virtual world.

[zalo]

Why hadn't I heard of this before today? It's a pretty cunning solution to the free movement dilemma.

However it's only a partial solution. Your head has a great range of motion between your stomach and your head. Latency from iPhone AR marker tracking has to be enormous. It's more like an indoor RedRovr.

It's cheap (for people with smartphones), it's creative, and it almost works. I'd like to see them do a redirected walking system with this to prove it works well. I also suspect that occlusion and robustness are a huge issue.

Perhaps they can mount the iPhone to the top of the head and paste the AR markers to the ceiling. Latency will still be ginormous, in addition to using a backtop or wireless system.

It's cool, but it's definitely not "the future". I can't see a chest strap and a floor strewn with AR codes becoming ubiquitous.

[WiredEarp]

t's cool, but it's definitely not "the future". I can't see a chest strap and a floor strewn with AR codes becoming ubiquitous.

It would be a lot cheaper than other tracking based systems, IF it can provide decent 6DOF tracking, not just of the users location, but of the gun etc. It looks as though they are working on that however, and if you read down they talk about the latency concerns. Id rather see it done with some dedicated hardware however to cut down latency, despite the fact that positional latency can be higher than head.

Really, whoever comes up with an accurate, reliable and fast indoor 6DOF positioning system, that handle at least 40 simultaneous sensors, will make millions.

[colocolo]

nice.

[bobjwatts]

I backed it. I think it's a good simple idea.

[Nighteyes]

Instantly turned off to the idea when I read it was iPhone only.

[shole]

I asked a few questions:

Is user height also tracked?
Yes
Android version?
No, I'll add an FAQ, but Android is not a good place to start with this because of the number of lens/sensor combinations.
Will software be open or proprietary?
Proprietary
Possibility to emulate x360/hydra thumbstick?
Unlikely, you'd run out of space fast.
Redirected walking?
Depends on the game designer, but redirected walking is difficult to pull off in an action game setting, but maybe not in an adventure game.
Possibility to strap it to my backpack with other hardware rather than chest?
You're the first one to figure that out! Yes, it can go on your back, the offset just has to be changed, and it frees up your hands.

I think being proprietary and requiring custom integration is the biggest issue with me supporting this.
I really doubt this will be "The Way" to do positional tracking in the future and this will end up as a development dead-end.
As an open platform it could be a viable stepping stone to bigger, better things.
We could hook it up to existing tech like emulating analog sticks to do the movement for us, putting it on better standing than the Omni.

But even if we had that I don't think it's a good way to walk in VR.
You would need a ridiculous number of tracking sheets and a lot of time to set it up.
Freewalking VR should be fast to set up for a session in a space you have limited reservation for.
We could still use a controller for actual walking and use this for tracking other bodily movement like stances, leans, side steps etc.
I would pay for that.
But I'm guessing the Sixense stems are better candidates for this anyway.

[shole]

http://www.kicktraq.com/projects/623678 ... made-real/
oh wow
i thought it was at least going to succeed, given its relatively lowly goal

[Benjimoron]

This is cool, but, notice how they cut the videos as the player is about to walk into a wall! I can't see how you'd be able to explore anything bigger than your lounge with this? Sure go and play on a basketball court, cool idea if you happen to own your own basketball court!

Why do we need this if we can use ODT's?

[cybereality]

iPhone only!!?!?!? I'll pass.

Guess us Androids aren't welcome here.

[android78]

iPhone only!!?!?!? I'll pass.

Guess us Androids aren't welcome here.

Gee, thanks mate! ... But I use an iphone! Oh, the irony. (BTW, my alias is one I've carried since I was in high school circa 1995 and well before android became an OS)
I think they'd be crazy to have this only for iphone since it should be a pretty simple port to android. Also, with the number of cheap android phones out there with good enough cameras (some of us have old android phones doing nothing), I could see some having a dedicated android phone just for VR, but can't see many would do that with an iphone.

[brantlew]

I think it looks cool and I will probably back it. Not sure why people seem a bit down on it. The motion tracking is way better than Red Rovr (although not as well suited for large environment). Although like Red Rovr, it's probably only suited for specific types of VR experiences and not general gaming as they are implying. Still...it's literally the only thing within a reasonable budget on the horizon that would allow you to experience free motion VR. And no matter how complex you make an omni-directional treadmill - it can never match the realism and generality of natural walking. It would be rather trivial for these guys to add redirection to the system and then it would be really interesting.

I do see some drawbacks. The camera system they are using has some severe limitations and I suspect that you must maintain a very rigid posture and cautious motion to keep the camera system optimally positioned and tracking smoothly. Also, the iPhone-only support is a pretty bad decision. While I understand that it is convenient for them to support a homogenous piece of hardware, iOS is just a horrible development platform, and iPhone is being trounced in the market these days - especially among the tech crowd.

[FingerFlinger]

Now, as for easy setup, a markerless visual odometry system coupled with the phone's GPS could be interesting.

[brantlew]

Now, as for easy setup, a markerless visual odometry system coupled with the phone's GPS could be interesting.

Yeah, but I suspect that you would still need to salt the environment to achieve decent performance. A good chance you would still need to throw down paper on the floor, but without the additional mapping step. It sucks that flat featureless environments for moving around in is exactly the worst-case for computer vision. An upwards facing camera and ceiling clutter would probably be best but difficult to create.

[FingerFlinger]

Yeah, definitely issues with it. A grassy field would probably be out of the question, for instance, but certain environments could work well. It's essentially how the Mars rovers(and many other robots) navigate, minus the GPS. Here's a JPL paper that talks about environmental challenges.

[shole]

I have a hard time seeing the markers on the floor thing working out.
It should be markerless to allow any useful application in outdoor redirected walking.

Most of captured video of markerless ground will surely be useless but at times when it reaches a certain confidence level it would be the most accurate sensor in the device, however rare.
So wouldn't it be useful to use the accelerometers as the primary sensor and use the camera to just verify the truth of those values?
I mean, the accelerometer has some hunch of where it was moving, but camera could get intermittent snapshots like moving at speed A to direction B or standing still and correct the estimation accordingly.

[Fredz]

If using a backtop it could be a good idea to use a PS Eye connected to it instead of an iPhone via WiFi. The frequency is much better (640x480@60Hz vs 1920x1080@30Hz) and it's a lot less expensive ($15 vs $549).

[Benjimoron]

If using a backtop it could be a good idea to use a PS Eye connected to it instead of an iPhone via WiFi. The frequency is much better (640x480@60Hz vs 1920x1080@30Hz) and it's a lot less expensive ($15 vs $549).

But the lack of pixels means that you won't be able to see the images clearly.

[FingerFlinger]

VGA is probably good enough for this application. I would be pretty surprised if the Atlas software was actually using HD resolution for their image processing; many real-time vision algorithms would downsample even further than VGA.

[Flassan]

I love what these guys are trying to do and with encouragement I think they will only do better. What stands out for me is that they are offering us an example of VR navigation using computer vision to try out for $30! I don't own an iPhone but I'm sure I could borrow one for long enough to try this out.
What I find refreshing is that I believe they are being honest. Personally I don't see much in the way of sharp editing, their demos with non-team members appear genuine and they aren't making unsubstantiated medical claims. They are selling the product, rather than the VR dream, so I think they making a positive contribution to VR.
The way I see it is if you learn anything from it, especially if it's unexpected like something you thought would or wouldn't work, then $30 is a snap.
I agree that the full works including international shipping may be higher than you want to pay, rather than too expensive. I'm less certain why they need to raise that amount seeing as it looks like they've done the right thing and proved it before offering it but I guess they have their reasons.
If you're interested in VR you might like to consider giving these Rifters a break?

[Fredz]

What stands out for me is that they are offering us an example of VR navigation using computer vision to try out for $30!

$30 is for the lite version without the chest mount and the printed markers, it's $100 for the recommended pledge. Add the cost of a iPhone 4S and you're at $550 for something that could cost less than $50 with a phone you already own. Nice effort and good idea anyway, but I wish it was a little bit more versatile.

[JustJoshin]

Whilst I think it is an interesting idea I just don't see it as being that practical. Let's face it, for the majority of gaming and entertainment experiences most people just don't have a lot of space to set aside. The footprint of something like the Omni is probably about the most people can set aside.

I think it is really exciting that we are seeing all these different ideas and people willing to work on alternate input and locomotion schemes for VR. I just don't think this one is practical for the average user in its current form / implementation. The iphone only is the least of its issues at this point.

[airasmussen]

@Fredz, we've actually messed with the PS Eye. The problem is you also need an IMU. Optical tracking isn't quite enough on its own. The IMU in the Rift could be used, but the simplicity of running the whole thing off a smartphone makes it a bit easier.

[Fredz]

I don't see how it would be simpler to implement an external iOS app to do that instead of using the PS Eye on the same PC.

Several people have already mixed positional tracking with the rotational tracking of the Rift with good success with the Razer Hydra (HydraDeck, Heli-Hell, Horrot, Titans of Space, etc.) or the PS Move (http://www.youtube.com/watch?v=7-rcU8cfdD4). The benefits in much lower cost, lower latency and higher frequency seem also quite important to me to not dismiss such a solution.

[JDuncan]

I made a theory then three concept ideas that use the theory, here;
http://www.mtbs3d.com/phpbb/viewtopic.php?f=138&t=18346

The third idea is relevant to the atlas kickstarter.
The third idea describes using a camera to find something on the ground, like the atlas kickstarter does, but my idea shows this is for static reference only.
There is a dynamic reference too.