VIRTSIM is like Project Holodeck on steroids

[benz145]

So one of the biggest US defense contractors put together a system called VIRTSIM which is a pretty comprehensive virtual reality platform. The only problem is that it isn't for entertainment purposes. Instead it is designed for military and other emergency personnel training. Why we can't go to our nearest theme park and pay $20 for this experience is beyond me!

http://www.roadtovr.com/2012/11/04/virt ... tform-1728

[cybereality]

Looks pretty cool, but clearly outside the budget of most hobbyists.

Could be great if they could retro-fit a laser-tag arena with those components.

[brantlew]

Very cool. I believe that everything here is within reach of the hobbyist community - not a few hundred bucks. But maybe a few thousand bucks - equipment wise. The gaping hole is the software. But an organized effort could supply that. It would be interesting if there was a massive open-source project to tie all the hardware together to create an experience like this. Of course the temptation to commercialize it into "laser tag" is so strong that it might be difficult to get enough people to "donate" their expertise for free.

[Okta]

Lol at their z800 legacy goggles But cool none the less. From what i can see even on these forums we are not to far away from budget setups that can achieve this outcome.
Raytheon do some awesome things though, who remembers the exosuit?

http://www.youtube.com/watch?v=-UpxsrlLbpU

[WiredEarp]

Yep have heard of VIRTSIM before (it was on engadget a year or so back as well)... its pretty awesome from all accounts. Shame about the FOV though! Once those guys get a RIFT they will laugh at what they thought was immersive before.

Brantlew, I agree with you. After reading all the discussions, it seems that optical is the best bet for achieving our tracking aims (multiple sensors in a larger area than magnetic tracking can cover).
That means that really it is just a software, calibration, and design issue. The hardware could just be standard PC's, and the cameras high quality webcams. If I was writing it, I'd probably start with a central server managing hubs which are processing the cameras, so the central server would just be reading data streams and crunching them, and then outputting its own stream to another server/process that was actually running the game.

The design aspect of the sensors is pretty interesting. I've seem some very interesting ideas on here about how to identify individual tags. I've thought that it might be also worth investigating using color to identify the individual people as well... That could be an easy solution.

[grandmaster789]

This kind of looks like one of the products I'm involved with (http://www.re-lion.com/products/suit/ta ... ur-of-suit), although it looks like they are using reflective markers and optical full-body tracking. We use a hybrid radio triangulation and intertial sensor system. It's very expensive and not intended for the general public, but the tech does exist already

edit - looking back at the movie clip reminds me that we really need a new one. This one is over two years old already, and the current version looks much better...

[brantlew]

@grandmaster789: Interesting. What resolution can you achieve with radio triangulation? The stuff I have read (at least for indoor WiFi triangulation) seems only accurate to several meters.

[RoadKillGrill]

I would think an inertial plus optical/rf would be much better than an optical system for this purpose. If there is multiple actors on the stage occlusion becomes a big issue, even worse when the output is realtime. Then again I'm a little biased since I work with inertial systems :p. Whats in the backpack? Is the rendering done on the user?

What budget would you think a hobbyist could pick this up at? Would $1500 for a single suit be too much?

[grandmaster789]

The accuracy from the radio transmitters we use is at best 10-20 centimeters, but it's pretty unreliable and tends to drift around. The inertial sensors drift a lot as well, but when combined the measurements from both systems improve each other by a lot.

The backpack contains a ruggedized computer, basically it's a custom backtop with hot-swappable batteries and lots of connections to wire up the inertial sensors. There is a single server that mostly handles communication between various 'soldier', 'observer' and 'operator' clients, while rendering is done on the backtop.

The main advantage of this system is really portability - setting up in just about any room can be done in about an hour. There are no real requirements to the room as well - there's no need to attach stuff to the ceiling.

I'm afraid the system is really far too expensive for enthusiasts. I'm not sure what the exact price is at the moment, but I do know that a 12-suit training system with all the bells and whistles costs over a million euros. Even just the HMD we're currently using is like $10K, so it's definitely not for home users...

I'll let you guys know when we've made a new clip with the current version.

[cybereality]

@grandmaster789: Sounds interesting. Would definitely love to see a video when you get a chance.

[grandmaster789]

We've recently made a newer clip, you can find it here: http://www.re-lion.com/products/suit/ta ... ur-of-suit

[Krenzo]

Good video. What tracking system are you using? Is it from Xsens? I'm surprised your RF inaccuracy is so high.

[cybereality]

Cool, man. The weapons looked good.

[android78]

@grandmaster789: That's looking really good. I like the complete system approach, with the software for building environments as well. Given this looks aimed at military, I wonder if you could get better results in terms of graphics, etc by licensing the crysis engine.
I'd also be interested in seeing how well this would work with the Rift... I can imagine that a dedicated enthusiast could get 8 rifts and have 4-on-4 death matches in a warehouse with a similar setup.
The 10-20 centimeters accuracy of the radio would worry me somewhat. I'm sure there must be some system that is more accurate then that, especially if it's as expensive as you have hinted at.

[grandmaster789]

We're using xsens sensors for skeleton capturing, with separate sensors for the guns themselves. Those work pretty good, but are still subject to drift so we're also using an RF system to compensate for that. None of the sensors are good enough by themselves, but taken together the worst of the measurements are compensated for. This is also for us an active area of research however (sensor fusion).

The system we're using has been in development for a really long time now, it was started as a research project a little over 10 years ago; we're continually looking for better solutions, and we're improving many aspects of the system.

As before, I'll let you guys know when we've got something new to show