Software Scoring Machine

[jBirch]

Hey JS,

Are you going to Open Source the code so we can all poke at it? I'd love to give a go at thinking about possibly porting the code to RT Linux.

James.

[JS Hart]

Quite frankly, my code is horrible as it is a product of evolution rather an intelligent creation... and I'm embarrassed to do that until I go through it and clean it up. But I do intend to do that eventually. I am hoping that the hardware side will get vetted, and I'll have a couple of iterations of bug fixing and code-improvement first. I realize that as soon as I open source the code I'll lose complete control of it, so I may open source a version that is without some of the bells and whistles and retain the code to the 'deluxe' version... I dunno....

The code is written in a very easy to use gamers basic called GLBasic, it's also very fast and has all the graphics and sound built in from OpenGL, and more importantly it has a low level timer function.

Anybody could write one, I'm a case in point (my degree is in Psychology). The language detects the fencer's weapons data by calling the gamepad functions... as easy as that.

All I did was write a main loop that gets the weapon data, and checks the time (to the microsecond). When a "touch" of gamepad data has been detected for enough milliseconds, a "valid touch" or "off target" touch flag is set, then the rest of the loop does the logic for lock-out, reset, etc , and checks for user command input, updates the screen (only when necessary) makes annoying buzzing sounds....

and then it's a box.

[JS Hart]

Oh, yeah...

and GLBasic cross-compiles to Linux, Mac OS X, and Windows CE, all-ready built in to the compiler!

[jBirch]

Ok, so I'm a professional software engineer and I've always dismissed GLBasic as a monkey language. I figured you'd hacked out a driver in C for Windows and had all sorts of low level timing problems.

I seem to be horribly, horribly, mistaken.

*grin*

Awesome work man.

Nobody pays for software, by the way. With all the open source stuff out there, it's a wonder us professionals still have work.

People will only pay for the hardware to run it on (a box with blinky lights) or for the service to actually install and customise the software. If you're looking to turn a buck, I'd suggest open sourcing the app and close sourcing the hardware (ie// get rid of the connector specs after I've had a chance to download them). Then, put the hardware in a box with a blinky blue LED and rounded corners. If you want to be REALLY slick, put a magnet on the back.

At any rate, there is no such thing as open-source clean code. It's all a hash. Just release it under GPL and we'll make it awesome.

Hope this helps.

James.

[Joe biebel]

I have real feedback for JS Hart's VSM. Shawn was nice enough to loan me an interface device that he made. I'm seriously thinking of claiming I never got it, so that I don't have to return it. It works incredibly well. I hooked it up to my laptop and it was good. I hooked my laptop up to my 22" office LCD, and it was awesome. I can't wait to attach it to my 46" LCD monitor!!!

Everything worked extremely well with no "dinking" around. After reading the instructions on how to build the interface, I was more than a little intimidated. After seeing it work, taking the cover off of it and seeing the insides, I'm enthusiastic again. I haven't built anything electronic ever. Now I'm sure I can do it. I would still be willing to pay someone to make it for me, but if no one wants to take it on, I will do it myself. Since I will be needing at least 4 of them, perhaps I will learn enough on the first one to be relatively quick on the repetitions.

Looking in Dell's "Outlet" today, I found a refurbished, warranteed, "slim tower" computer that will be 10 times what I need, running Windows XP for $289.00. I have numerous monitors available, so my final tally will be about $370.00 per system. After seeing it work and playing with it for a day, it is IMHO superior in every way to a scoring machine. The visibilty and features are the best. Your experience will vary depending on the display you use.

One of the great things about this system, is it's upgradeability on the hardware (monitor) end. In about 2 months, when my new fencing club is up and running and I have at least one of these systems installed, I will post some video, unless someone beats me to it.

Shawn, thanks so much for making this happen and a special thanks for the loaner.

[SenorEste]

Wow.

The idea of a software machine is great, and what you've done is totally awesome and elegant. Open sourcing this could be a great thing for the fencing community, and as others have mentioned has very much potential for a large impact on the sport as a whole. Giving clubs and teams access to cheap, customizable, and configurable technology will only help the sport. The companies that make the boxes will also be forced to adapt (in a good way) if this thing gets popular at all.

In it's current state, it seems most impressive. Imagine the possibilities though! Built in pool/DE assignments that could call/notify fencers for you, networking to a central computer to calculate and post results on the fly, locally, and onto things like FRED. Ability to change/edit timings to train/experiment with rules changes. Attach a web cam, and we could get instant replay, video output, and snapshots of touches. Built in troubleshooting that tells you maybe why your weapon is not working, for clubs/people new to that stuff. BUILT IN RULE BOOK that could be called/searched for specific terms for the director to read/explain!!! Make it run in linux and other OS's to reduce costs even further.

Features like that, and more could be truly revolutionary.

[JS Hart]

Thanks Joe...

There will be updates and bug fixes for a while.

Saber needs work the most as it was the most difficult to make work from the hardware point of view, and still has problems with spurious hits appearing when both saberists are intermittently grounding their weapons to their lames (which they tend to do a lot) and it also seems to miss some hits, especially if they are "through steel". This appears to be an artifact of rapidly changing electrical states and "transition" states that are mis-detected by the adapter. I think I can get around these problems by processing the data better...

Foil really seems to work quite well as long as the "sweaty fencer" effects aren't too bad. I intend to change the way the hits are detected in foil from a "rule-based" system to a simple data-lookup-table and enlarge the set of possible data sets to include the common sweaty-fencer data (which has not been consistent with my rule-based "perfect-world" algorithm.)

Epee is already data-based, and seems pretty bullet proof so far.

When I get that far, I'm leaning toward open-sourcing a generic version so the rest of you all can tinker...

-- Shawn

[Joe biebel]

Last night I took my laptop computer with Virtual scoring machine, my wireless gamepad and the VSM adapter to the Caliburn Fencing Club for a night of fencing. Wow! This is the way to score touches. Even with a relatively poor laptop and screen (measely 135 deg. viewing angle) everyone agreed that it was very easy to see. It was used for foil all night and I spent the night refrereeing and learning to use the "remote control". The large (compared to most scoring machines) face, lights, clock, and information displayed, were "eyepopping". There was not a single complaint about timing or functionality of the system.

Even when we were done, the conversation went to "what ifs" and what's next. This is a real breakthrough. Now, after refereeing with this system for several hours, I think it is incomparable to any existing machine. I would prefer it even to a Favero "fullarm" fencing computer for information display and features.

I think as more code is written, more features can and will be added to future iterations of Virtual Scoring Machine software.

One thing that happened several times, was that the machine switched to Epee, while the fencing was going on. This could have been "user error" with the remote. I would like to be able to disable this automatic weapon select feature that switches it to epee.

When fencers were done with fencing, and lessons, etc., many came over to ask questions and to "nit-pick" the features. The most common comment had to do with the logo and info tag that are displayed on the screen at all times. While it's nice to know the product information, perhaps a "splash" screen at startup would be a better way to go. Aslo, everyone liked the sound the machine makes to increment and decrement the touch count. Many funny comments and suggestions were made in this regard.

After Shawn sent me a loaner adapter to try out, naturally I got incredibly busy with work and doing the buildout on my new salle. At any rate, it was great to have this opportunity to try out this system. The next time I do this, I'll take video of the system working and post it on youtube or something. I swear, once you guys see this work, you'll have to have one.

[Joe biebel]

I'm sure I sound like a fanatic about this system so, I would like to invite any of the Caliburn Fencers that experienced this machine to give comments as well.

[migopod]

I didn't fence on it, but having watched some bouts I can confirm that the the 'lights' were very visible from just about all of the same angles and distances as a box with flush-mounted lights (eclipse, SG, etc). Also using the RF/BT/whatever game controller as a remote looked a great deal less awkward than using IR remotes. Presumably it also offers the possibility of playing some Quake or similar between bouts.

[SJCFU#2]

Joe,

Could you keep an eye on how the VSM responds to "irregular" equipment such as lames and body cords which might be fine for practice but would never pass inspection?

There are a lot of requirements related to resistance in Appendix B to the material rules and I am curious about how the VSM adapter compares to a dedicated machine (it may be more forgiving, it may be less - probably depends on how the optocouplers respond to changes to the input current). I've been thinking about building an adapter just to test this however it sounds like you are way ahead of me.

[Jvanhousen]

Originally Posted by Joe biebel

I'm sure I sound like a fanatic about this system

Just a bit, here's a pic of you w/ the controller:

[jBirch]

Great. Now you'll have them trying to figure out how to operate the box from your iPhone. Do we really need yet another reason to buy an iPhone??

And I'm wondering if inter-poule Quake will improve the quality of direction at the end of the day. Might be interesting to find out...

James.

[migopod]

Originally Posted by jBirch

And I'm wondering if inter-poule Quake will improve the quality of direction at the end of the day. Might be interesting to find out...

Attack from left with BFG9000 arrives.... on everybody nearby.

also picture all the epeeists camping by spawn points rocket jumping up and down.

[Joe biebel]

Originally Posted by SJCFU#2

Joe,

Could you keep an eye on how the VSM responds to "irregular" equipment such as lames and body cords which might be fine for practice but would never pass inspection?

There are a lot of requirements related to resistance in Appendix B to the material rules and I am curious about how the VSM adapter compares to a dedicated machine (it may be more forgiving, it may be less - probably depends on how the optocouplers respond to changes to the input current). I've been thinking about building an adapter just to test this however it sounds like you are way ahead of me.

One interesting thing that I noticed was that the "ground" contact lights appear when you touch your opponent in foil with the tip, but do not depress the tip. I think this could have been developed or is a byproduct for the special features that indicate sequence, timing and debounce timing. I used it to show that a lame was bad and where it was bad, without setting off the lights and waiting for a reset. You can just move the tip around on the lame and watch to see if the yellow light disappears.

When you say "it sounds like you are way ahead of me", you are really talking to JS Hart. I am only using the hardware and software he developed, to assess and to report. After I get over the novelty of the system, I will explore in detail the "special" features like playback and start looking at the timing and debounce info.

[sir alasdair]

Ok,

So here's my opinion. I will attempt to look past the novelty of the idea and just say what I thought of the machine from a practical standpoint.

As a fencer -

I fenced 4 bouts on the machine on Wednesday night and found it to be an efficient and useful tool. The display - even on a 15" laptop monitor - was easy to read and understand. I had a few instances where I felt some touches did not register that should have, BUT that may have been me just not exercising my normally PERFECT point control. As Biebel pointed out, he has not used the replay feature yet, so once that is an option, one may be able to see if a touch was recorded, but not long enough to score. From a coaching standpoint, one will be able to better discern whether a touch landed flat, or landed true but was too quick, and give you the knowledge to correct it. All in all, this system should be the predecessor to the future of scoring machines. I would hope that the software can be written for Linux and Apple as well to become more universal - besides, Bill Gates has enough money.

As a Referee...

Ok, I didn't referee with the unit, but I stood back and observed the unit during my down time. The advantages I noticed early on were first of all the remote control unit (wireless video game controller) was comfortable in the hand and worked from all angles, even the far ends of the strip. I have refereed with the favero machine with the remote and had a hard time getting the clock to start from the ends of the strip. The primary disadvantage I can see from a refereeing standpoint is the learning curve with the controller and software for directors who are not savvy with video games or computers. I have directed at tournaments and been handed a remote with NO prior knowledge on how to use the equipment and had to ask or figure out what do do. None of my club's machines have scoreboards or timers on them. It becomes hard to focus on directing fencing if you are worrying about working the machine. This unit requires a prior knowledge to use and could really present problems for "on the spot" training.

I couldn't help but wonder if there was a combination of buttons that allowed one to make the fencers jump up, grab a hammer from above and smash the opposing fencer over the head ala' Donkey Kong.

Overall Rating: Thumbs Up

OP ED: This type of system should be the next step in the evolution of fencing. Will it be? That depends on the FIE and their infinite wisdom. There will likely be pressures from vendors and equipment manufacturers of current systems as this would affect their business.

[finnfence]

How much were you thinking of making from this system? Maybe if we all donate a bit to bring you up to what you wanted for it, then you could release it for free, and earn the eternal gratitude of student fencers everywhere....

[griffindm]

Originally Posted by finnfence

How much were you thinking of making from this system? Maybe if we all donate a bit to bring you up to what you wanted for it, then you could release it for free, and earn the eternal gratitude of student fencers everywhere....

Finnfence,

He IS releasing it for free, in the basic version most users will be interested in. He is just not releasing the source code. He has said a version of the software source code without the advanced features, which relate mostly to timing and history, may be released in the future. Even the version with the advanced features is only US$10.

Dave G.

[Dead6re]

Has anyone figured out how to use a standard USB cable with the patch and link it directly to the adapter?

Also, for those that are using USB, can you tell me what your kit list for making an adapter is?

I'm very interested in setting up a Linux computer and porting software to C++ to get it running but I need some basic starting points?

[jabalino]

OK, so I finally built a working copy of Hart's box with a couple of differences. First, I built mine without sabre capabilty and without the capability to hook up a grounded switch. Awesome ideas, but I'll add them in later. I soldered it all up without the circuit board. It takes a little longer to wire it that way, and if I had to do it again, I'd get one made since the plans are there and it's cheap to do.

The P220 gamepad is pretty easy to take apart. Definitely test it first. I didn't and found out that the one I bought was broken from the start (windows wouldn't recognize it). A second gamepad worked fine right from the beginning. Instead of cutting the pad with a dremel, you can find soldering points already on the board to tie into, and I recommend doing that. Just trace the connections back, and you'll see a small hole that you can solder a wire to directly. Saves a little time and anguish in my opinion.

I've integrated these connections into my java software, so version 3.0 of my java software should be fully compatible with Hart's design, as long as the P220 or keywiz GP is used. As a reminder, the keywiz GP will work fine, but it doesn't go any faster than 8ms ticks. It's a little easier to wire up, but also a little more expensive, and flicks in foil will always be a little off. I strongly recommend the P220 route, but I'm leaving the keywiz GP supported for now anyway.

The current version of the software is compressed in a jar file. You'll need these three files. You can replace the buzzer.wav file with whatever you like to use as the buzzer (as long as it's in wav format, obviously). Since it's java, it should run pretty much anywhere. I'll be testing it on linux soon, and I'll let you know how that goes. Keep in mind that (as with Hart's software) windows by default keeps the polling rate at 8ms. To boost that to 1ms (i.e., to get the timing fine enough to register flicks correctly in foil), you need to patch the usbport.sys file, and that entails some risk for the stability to your system. One site describing this procedure is (http://www.ngohq.com/news/12163-how-...ows-vista.html) and Hart includes some software to do it on XP. I'm hoping that some versions of linux have a high sample rate automatically, but I'll look into that. If you don't care about the correctness of flicks, you can skip this step.

http://www.geog.ucsb.edu/~pingel/Fen...ncingBox30.jar
http://www.geog.ucsb.edu/~pingel/FencingBox/jxinput.jar
http://www.geog.ucsb.edu/~pingel/FencingBox/buzzer.wav

So, far all practical purposes of foil and epee, the hardware is there and it can be built very cheaply (certainly under $50) and with a moderate level of soldering skill. Now, I'll continue to develop the java software. I'd like to see the software side go open source java so we can continue to develop features and find good ways to integrate more capabilities of the laptop into the tournament. For instance, I'm going to try to get pool sheets and DE tables integrated into the scoring box (think something like a "practice mode" and a "tournament mode") so that the box can get the sheets from the bout commiteee computer and relay them back when finished. This is more of a proof of concept enterprise than anything else. I'm also planning to integrate a few more handy buttons on the interface to make mode switching, volume, etc, a little more convenient than the menu bar. Ideally, I'd like to integrate this kind of thing with touch-screen capability.

Anyway, so the upshot is that if people want to build Hart's box and use with linux or on a mac, they now can (albeit with some more limited features).

-Tom