Live Fire Half-life

We don’t know how we missed this when it first came out, but there is a hack out there that combines a .22 caliber pistol with the video game Half-life. Simple is best and that motto is in use here. A wall was built down range to use as a projection screen. Accelerometers mounted on the drywall report vibration data from the bullet strike which is used to triangulate its location. This targeting data is then sent to the game interface.

As you can see in the video after the break this works like a charm. The [Waterloo Labs] personnel that developed this are also responsible for that iPhone controlled car. The antics we witnessed in that project carry over to this one as they illustrate using the setup to play Half-life with a couple of shovels at 2:12 into the clip.[youtube=]

[Thanks Entropy]

48 thoughts on “Live Fire Half-life

  1. Well, “real” is a matter of definitions here. They used a suppressed .22 with a long bull barrel. Not quite like shooting a Glock Model 22 or a 1911.

  2. How exactly is it a waste of ammunition? People go to the gun range every day and shoot hundreds of rounds at paper targets. This doesn’t seem to be any more a waste of ammunition than target shooting, unless you’re hoarding it for the zombie apocalypse. And, they’re shooting .22lr, which you can get 500 rounds of for between $15 and $20, depending how much ammo the teabaggers are buying up to hoard.

  3. Wow. I’m sorry to say that this is the first time I’ve seen something like this on hack-a-day after seeing it somewhere else, a few months ago.

    Also, any good practice is no waste, especially when it is just some el-cheapo .22.

    Although, when you consider that they were having more fun with the shovels, maybe it is a slight waste…

    1. The army has something like this that they use for some sim games. I’ve seen it first hand. That was back in 2005 I think. It is a trailer with a ballistic wall in the back. The game is a rail shooter. I need to find links.

  4. That is pretty cool.

    Fully silenced gun? Must do research.

    Seemed to have a 1 second lag time, too much bloat in the software? If Nintendo was doing a better job with the PowerGlove, they can do better.

  5. As some have pointed out they are not triangulating.

    You could use two sensors and triangulate, would be easier to calculate.

    3 sensors provides 3D space, and on a flat board seems redundant.

    They didn’t give any clue what their “rock hard” surface was, but since it appeared to be sheetrock I don’t think it was rock hard, you can punch a hole in it with ease.

  6. That’s pretty damn cool…I’ve been thinking about designing some pop-up targets, but this has me beat by a mile.

    And Caleb, I’d love to see a link to that. I ran the electronic ranges when I was in, but the weapons we used, while they looked (and functioned) exactly like the real thing, were air powered, and had an IR laser in the barrel. We even had M240s, AT-4s and shotguns to play with. I’ve never seen anything similar with live fire stuff though.

  7. It is ridiculous that they go “Go ahead and try it yourself and have fun with it!” OK, Lets recap what we need right:
    1- 1x LabView $1500 basic version
    2- 1x USB Data Acquisition Device NI USB-4432 : $2200
    3- 4x 50 mv/g accelerometers about $200-$300 each.

    Total project cost w/o computer, projector etc: $4500

    And you guys are worried about ammunition. :)
    Yeah, I’ll run and try this right away…

  8. Next week: How to build this rig using three modified piezo buzzer transducers + 16 MHz AVR microcontroller.

    What’s needed:
    – dimensional drawing for custom piezo transducer mountings and glued-on center-mass

    – schematic for instrumentation amplifier/filter circuit and three-channel (fast) ADC board

    – AVR (or arduino-) firmware

    – PC interface (gameport or USB)

    – calibration instructions

    Could be done…

  9. @nubie:

    3 sensors are required for triangulation (Hence the term “Tri”angulation). Using total field sensors (e.g. sensors that lack directionality) In 2 or 3 dimensions, 1 sensor will limit the search space to a circle for 2D, and a sphere for 3D. 2 sensors will limit the search space to 2 possible points (one real one imaginary) in 2D, and a circle in 3D. The 3rd sensor will give you a single point as the location of interest in 2D, and a search space of 2 possible points (one real, one imaginary) in 3D. A 4th sensor will give you a single location of interest in 3D, however this is generally not necessary (Depending on the application) as the imaginary point in 3D is often easily distinguished by constraints imposed on the search space.

    Most sensors however do report directionality (+/-), but are sensitive only along a single axis. Most “sensor packages” make up for this by including multiple single axis directional sensors arranged such that their sensitive axises are in an orthogonal configuration. So a 2 axis sensor package is not 1 sensor, it is actually 2 sensors aligned orthogonally.

    Using 2-axis sensor packages, you can then triangulate in 2D with 2 sensor packages (e.g. 4 sensors), assuming that the information you are sensing has components in more than one axis. This is because the gradient of the signal between the 2 sensor packages will tell you which of the 2 possible locations is real and which is imaginary. For 3D triangulation you still need a 3rd sensor package and the packages must be 3 axis sensor packages (Again the gradient of the field will eliminate the imaginary point for you).

    With that background information, now consider the situation at hand. Since they are using accelerometers on what we can consider to be a flat plane, the information they are sensing will be transverse wave deformation of the flat plane, which is measured orthogonal to the plane.

    The 2 axises lying in the plane will contain very small high frequency longitudinal wave signals that are unlikely to be above the noise floor of the sensors, which means that unless you plan on machining precision mounts and spending a great deal of time calibrating the system in order to put the transverse wave signal into more than one axis of a single multi-axis sensor package, you will need to use 3 single axis sensors at different locations in the plane. Also, the calibration for the complicated 2 sensor package multi-axis setup will be rendered void after a few shots as small semi-permanent angular deformations will translate into significant signal being moved from one axis to another. This would give you wrong and or impossible answers.

  10. Correction:

    The “Tri” in triangulation refers not to the number of measurements, but just to the fact that you are calculating a 3rd position from 2 known positions and 2 angles. This also makes more sense in the context of the rest of my treatment of the subject.

    Also, I am in agreement that what they actually used was a Multilateration technique. Actual triangulation could still be done by comparing phase between sensors, but you still need 3 sensors to resolve the real point since a single axis sensor alone will give you 2 possible angles.

  11. Yea, the price here seems quite ridiculous and extremely laggy as concino said, but this could easily be done with an AVR or PIC and maybe an fdic chip… (as Ulrich noted).

    I think some more thought could have gone into the project, rather than using all the most expensive equipment they could get their hands on.

    Maybe something even more fun would be getting a laser pointer, strapping it to a laser tag gun, and goin’ at the wall at home. Any ideas on how to get coordinates of a laser tag gun? I’m thinking something like using an IR sensor (like in the wiimote) would suffice.

  12. @t:

    I was speaking more to the claim that triangulation could be done with 2 sensors than talking about what was actually done. They could have used triangulation techniques but used time difference of arrival instead. I did fail on the trying to be clever with the “Tri” thing tough…

  13. If the 2 sensors were at or near the edge of the board then there will be only one correct location, not 2.

    I just see no need for 3 sensors to detect movement in a single plane constrained to a square.

    Why do these guys have funding, and why are they taking the fun out of funding?

  14. I made a little sketch of how multilateration could be done with only two sensors:

    As you can see, the sensors give two possible positions, but one is clearly invalid.

    This could be done with pure triangulation instead, but it would be much more complex. To triangulate, you need a pair of 2-axis accelerometers. By measuring the X and Y magnitude in each accelerometer, you can calculate a direction vector. Two vectors originating from known positions yields the hit point.

    To further extend this project, use an impermeable backdrop instead of sheetrock or concrete board. Commercial installations use a flexible cloth impregnated with steel fibres. Solid steel could be used too, if angled to prevent ricochets.

    I see no reason why this couldn’t be done with a pair of piezos (with bulletproof shielding, of course) and a microcontroller with HID software. This doesn’t really lend itself to first person shooters, but there are plenty of arcade rail shooters this would be perfect for. And Duck Hunt.

    I volunteer to help create this. I’ve been looking for an excuse to get a gun, and this seems perfect. Contact me if you’re interested.

  15. I’d be more tempted to blame the flash game than the hardware for the lag :)

    For those complaining about the waste of ammo, you could play this game with a pile of rocks and strengthen your throwing arm, or as demonstrated get some crowbar practise for the real zombie apocalypse. Alternatively, consider it a proof of concept for better target ranges for training the military, police, hunters etc.

  16. I doubt you really need one of those expensive accelerometers to do this. All you need is some signal when the acoustic wave hits the sensor. You could probably get away with a simple piezoelectric buzzer (dirt cheap from electronic goldmine: ). Also, as stated previously, the Labview software could be replaced with a cheap DSP to perform the calculation. In fact, you could probably build a mouse replacement easy enough.

    First one to build a USB plug-n-play version for under $100 gets mad props.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.