Software Scoring Machine

[jjefferies]

Originally Posted by griffindm

JJefferies,

The British Fencing Association has the "best" english translation of the current FIE rules. You can find them here.

There are no substantive differences from the USFA rules as far as timings currently.
Dave

Dave, Many thanks.
Interesting what is specified and what is not. The language is more legalistic than technical - well I suppose it would be wouldn't it? It is all in Book 3, Material Rules in Sections .45 and Annex 3. It appears to be more about the certification process.

The terminology is definitely British. For instance the use of "accumulators" which I would interpret as capacitors in American English and "mains". One gets the main point that 12 volts is approved and no higher voltages. Definitely not written from the standpoint of microprocessors. But I think that one section approves the use of "encoded" wireless.

Ah, to bed and perchance to dream of scoring machine cheap and sturdy. Less than USD $50.

[jabalino]

Recent testing revealed a bug with epee that didn't throw out touches to the bellguard. My apologies! I had rewritten that section of code and bench tested with alligator clips between connections and not actual epees, and consequently missed something. Sometimes connections made with epees aren't quite the same as the "same" connections made with wires due to subtle differences in timing. Anyway, the software has been updated, and I've included a version number in the header. This release has been arbitrarily dubbed version 0.22. This _should_ make epee 100% correct logically, with any issues of timing to be determined. I'll be testing out the epee more tomorrow for a couple of hours.

Still haven't received the USB version. Great idea about the external lights! I'll think about that. I want to stay away from using a parallel port and stick to USB if I can. I'll look around to see what is available on the market that could be converted to an external light system.

Also, as far as external control goes, I'm finding that voice control can be a little too spotty, especially with starting and stopping the timer. I haven't tried the engine that comes with Vista yet - maybe it is better. In the mean time, I looked at a few presenter-mice online. It seems they pretty much all have controls for arrow keys pretty prominently displayed. I'm thinking I'll make the right arrow touch right, left arrow touch left, and the down arrow to start/stop the timer. The up arrow could act like a shift key or something, and if held down could remove touch left/right/reset the timer. I don't actually have one of these, so I won't really be able to test it, but it isn't hard to program in.

How convenient do you guys find the score to increment for epee automatically?

[oiuyt]

Originally Posted by jabalino

How convenient do you guys find the score to increment for epee automatically?

As long as the feature is disabled I don't mind it at all.

-B

[griffindm]

Given how often epeeist like to test their tips (and I am an epeeist myself), I'd have to agree with Brad.

With the presentation remotes, I think you need at least 8 keys for full control
i.e.:

1 + touch right
2 - touch right
3 + touch left
4 - touch left
5 start clock
6 stop clock
7 add time to clock
8 subtract time from clock

The last two being necessary to cover the additional priority minute as required. (unless you could just toggle between 3 and 1 minutes).

Make that last ones kind of hard so it doesn't happen by accident.

[brtech]

Presentation remotes don't have that many buttons, but you could use one for some kind of shift key.

The unit I referenced is a receiver for an IR remote control like any of your TV/video/audio/home theater systems use. They have lots (too many) buttons. The problem with IR is off axis response. Fencing machines use multiple (3 or more) transmitter LEDs and multiple sensors to improve the range and angle of the remote.

[oiuyt]

Originally Posted by griffindm

Given how often epeeist like to test their tips (and I am an epeeist myself), I'd have to agree with Brad.

That's usually helped by the auto-increment only happening for touches scored while the clock is running.

The issues I've seen are when the clock isn't properly started or stopped, or when a non-valid touch is scored around the time the stop button is pressed. Now the touch could increment or not, depending on which happened first (blinking score for a period after the increment helps with this).

I've seen way too many touches not get recorded and way too many touches get recorded erronenously to be at all interested in having this "feature" on a scoring machine.

As a referee I'd rather not have the scoring machine trying to think for me on awarding touches. It's really not that hard to have to hit the button for each touch, and I'd prefer to have control over something that important.

-B

[mrbiggs]

On an unrelated note, the increment and decrement time thing is a necessary feature omitted in most S&T boxes I've used. It's not often needed, but there are times.

For example, if the referee says "halt" and the remote is not working quite right, he can look at the time rather than running over and frantically pressing the button, while the bout loses more time. Or when the box runs out of power and the referee knows that there was 1:23 left, so everyone has to sit around and watch the box run for 37 seconds. Or when time runs out on the bout because the ref forgot to press the button, and has to estimate the time remaining (which was clearly close to 10 seconds and not 1 minute).

[griffindm]

Originally Posted by jabalino

Real time linux is the way to go, especially if you can put it on a bootable CD/DVD/ or thumbdrive. I think as a mitigation, though, the java app process could be assigned a real time priority in the windows task manager. Obviously not a total fix, but it may help in some cases. brtech - Yep, you could probably use a parallel port, but even fewer machines have those these days than PS/2. Looks like the Happ USB model that griffindm posted works at 1ms intervals. It's out of my price range for testing, though. If it can be done, though, the guys at keywiz will probably integrate it before too long, and for a much lower price.

First club test of the machine tonight, I'll let ya'll know how it goes.

Tom,

Regarding RT linux, I found the RTAI site here to be what appears to be both currently supported and open source. I think ultimately if we are ever going to provide a software platform for a fencing machine that can do everything that we get excited about (low cost, customizable features, timing, networking, etc, etc) and be fast enough to meet any proposed changes in the timing specs, then some type of low cost, bootable RT OS will be required.

The question is, do any of us here at Fnet have the time available and enough linux experience to make it happen? I don't, but I would love to hear from someone that does.

A second question is whether such a bootable config would be generic to run off most Intel-type machines, or if the kernals would have to be customized for different machines?

Dave

[brtech]

Originally Posted by griffindm

Tom,

Regarding RT linux, I found the RTAI site here to be what appears to be both currently supported and open source. I think ultimately if we are ever going to provide a software platform for a fencing machine that can do everything that we get excited about (low cost, customizable features, timing, networking, etc, etc) and be fast enough to meet any proposed changes in the timing specs, then some type of low cost, bootable RT OS will be required.

The question is, do any of us here at Fnet have the time available and enough linux experience to make it happen? I don't, but I would love to hear from someone that does.

A second question is whether such a bootable config would be generic to run off most Intel-type machines, or if the kernals would have to be customized for different machines?

Dave

I can answer the latter question. You can build a CDROM that would boot and run on nearly every machine. Probably, the app could run off of that, although if the computer had a disk, having the system build a bootable system on the disk would be best.

[griffindm]

Originally Posted by brtech

I can answer the latter question. You can build a CDROM that would boot and run on nearly every machine. Probably, the app could run off of that, although if the computer had a disk, having the system build a bootable system on the disk would be best.

Brtech,

Would that be equally possible with a USB thumb drive (assuming the BIOS allows booting from USB)?

The RTAI site seemed to infer that the RT side was only a couple of mods to other distros that would allow for a RT thread process. If so, Java support should not be a question. Just a matter of the right person with the proper knowledge working the platform issue.

[jabalino]

So my keywiz GP-Wiz MAX (http://groovygamegear.com/webstore/i...roducts_id=235) arrived yesterday, so I wired it up. Programming was a bit of a pain, as Java doesn't support game devices natively. But the work there is done, and we'll be testing the new unit out tonight at the club. Preliminary testing of flick shots indicates a slightly coarser resolution than the PS/2 unit is sometimes capable of, but (as advertised) it hovers at around 8ms ticks. I posted a list of the debounce times for the USB version under XP and Vista (they're pretty much the same) for a good round of flick shots to the carpet. The short story is some go, some don't, as you'd expect. The times are located at http://www.geog.ucsb.edu/~pingel/Fen...k_times_XP.txt and http://www.geog.ucsb.edu/~pingel/Fen...imes_vista.txt.

The gamepad-based hardware is nice because it doesn't send keystrokes to interfere with other applications when you aren't using the fencing box application, so the on-off switch is pretty much unneeded. I'll post a new wiring diagram for the keywiz GP unit and the software within the next couple of days.

BTW, during some research on real time process scheduling, I found this product (http://www.tenasys.com/products/intime.php) which is kind of interesting.

[griffindm]

Tom,

The Tenasys site looks promising. Just wish they had more open information on run-time pricing. The free download of the software does NOT include the USB RealTime driver, but it is included in the product.

Even if the timing of the OS can be resolved with this, I'm not sure the HID interface of USD will support the kinds of timings we would ideally want. Your timings you obtained with it look pretty good though.

I'm looking around for other USB I/O interface controllers. If they have both input and output a single controller could also handle repeater lights. How many inputs are you using on the keywiz GP-Wiz card?

I have an EE senior student on my college epee team, and I'm trying to interest him in the project to get his input as well.

Dave

[jabalino]

So far just 6 inputs, but I'm going to have to increase that as I try to solve some of the remaining connection issues. I've been thinking about looking for an old gamepad at one of the thrift stores in town, and seeing if the rumble effect can be converted to an external light system. As you say, we'd have the input and output in a single device that way.

The first club test of the USB system was a little disappointing. Flicks went off way lighter than they should have, and there were some other strange timing issues that didn't happen with the PS/2 version. Just guessing, but it seems like things are buffered differently. I'd hold off buying anything yet, although I will say that I think it won't be too hard to build in support for just about any gamepad device. I won't post the new code until I get some of this worked out a little better.

[brtech]

Yeah, it could work on a thumb drive

[griffindm]

Tom,

I've sent an inquiry off to Tenasys about the Intime development and runtime pricing. I'll let you know as soon as I hear something back. If we could get this to work on WinXP, the whole bootable OS issue might go away.

I also checked into other USB I/O control cards. The USBMicro U4x1 series runs $35US and gives you 16 lines individually configurable as input or output. The device bandwidth is 800 bytes per second as a HID device, which I believe translates into 1.25ms response. Information on it is here

Dave

[JS Hart]

I seem to have arrived late to this party.

I've had a working USB 3 weapon laptop scoring adapter for about a year. Saber is a little flakey (minimum 1 ms timing resolution is required), foil and epee are solid. The USB timing resolution problem was resolved by patching windows XP to update the HID class USB driver to poll every 1 ms (vs the 8ms as per the MS driver). Electrically, I used a hacked $10 gamepad as the starting point, and added optical buffers. I also incorporated a physical hardware switch to change between foil/epee and saber. There is a ground for the piste.

The software automatically switches between foil and epee when it sees what weapons are plugged in. I have "flick" displays for foil, and "double" displays for epee and saber (shows the relative timing of hits that "real" machines ignore.) I also added an "Instant replay" mode for reviewing what really happened and when (to the millisecond). Scoring and timing is integrated. 1, 2, or 3 periods of 3 minutes can be selected.

Current issues: it gets "weird" when people get too sweaty and conductive. I think this is an artifact of being too sensitive/high an input impedance. I'm thinking about adding a layer of simple transistors to buffer/control the input levels. Saber isn't fast enough, it can miss events or mistake them occasionally.

-- Shawn Hart

[griffindm]

Shawn,

Welcome to the party. It sounds like you have had considerable success thus far. Can you explain how you did the HID patch to XP? I was hoping we could find a way to get a generic Windows machine to give a fast enough sampling rate. Does your experience with any logging of debounce times indicate you are really getting close to 1ms in both the OS and the USB controller?

Please give us a link to look at your app. Are you doing this as a development for a commercial product or just as a demonstration/freeware type app?

Dave G

[JS Hart]

XP HID patch: search on "XP-USB-Patcher.exe"; it was a gamer's patch for better mouse speed/resolution I found a year ago. After I installed it there is another app (mouserate.exe) that verifies that the poll rate has changed. In addition to that, once installed, my "Virtual Scoring Machine" was then able to see "flicks" of 2, 3, 4, 5 ,or 6 milliseconds. If you can't find it email me directly and I'll send it to you. I found that the gamepad I hacked also had to be able to run at the faster poll rate -- and not all are.

My intentions thus far are to get the design stabilized and publish them with "basic" software free. "Advanced" software with the flicks and instant replay I was thinking of asking a modest $5 or $10 for... Is that reasonable?

I have not built a web page to link to, but if you really want to see my amateur attempt at engineering, email me and I'll send it to you.

I'm corresponding with Tom out here in SoCal and hoping we can cooperate on this project, and maybe fence and drink beer, too.

-- Shawn

[jabalino]

Shawn's stuff is pretty sweet. Hopefully I'll have some time to build a prototype next week or so and try it out, but the circuit design solves the problems that I hadn't worked out yet. I tried the XP usb patch - it works for my mouse, but didn't do anything for the keywiz USB adapter, so it looks like a hardware limitation there. I think the wiring could be pretty easily combined with the PS/2 version though, if someone didn't want to take apart a gamepad. I think gamepad is the way to go, though, especially if we can get the rumbling to act as external lights. Even without, it's just too handy not having to worry about extra keypresses when you're outside of the app.

BTW, for anyone playing along with the current version of my stuff, I updated the software with a new interface, streamlined the code a little bit and fixed another bug in epee and one in foil.

[griffindm]

Shawn,

Glad to hear you and Tom are corresponding. Any chance you could get the USB controller card info off the gamepad you successfully modded so that we can check its availability and cost. Can't be that expensive if it is used in a gampad, can it?

Tom,

Can you publish your timing results with the PS2 after the patch?