Recent Comments:
DIY robotic foosball table is ready to throw down {Engadget}
Dec 26th 2007 1:00PM I took apart the camera to look at the insides, to find a USB 1.1 chipset. USB 1.1 has a limited bandwidth, which prevents high frame rate/resolution combinations. The camera can do 30FPS, but only using compressed images, and not at 640x480. The windows driver must do a lot of smoke and mirrors with the image set, because its not phsically possible for the camera to do a higher combination than that.
DIY robotic foosball table is ready to throw down {Engadget}
Dec 26th 2007 11:49AM oops looks like I got cut off:
...we only had $500 and 3 months to work on it.
6) Yes, we realize there are other systems out there that do the same. But there is only 1 commercial system, and its sold for $27,000. There are some other universities that have developed some too, but I'm guessing they probably have more than a $500 budget and probably more than 4 undergrads working on it, not to mention more than a 3 month time frame between initial thought and final demonstration. It would have been great if we had a mechanical engineer on our team (or maybe a computer science major). At the same time, while our project probably didn't turn out as well because we were missing those guys, I learned a lot more personally, so I'm kind of glad. In the future, I realize why and where belt systems are to be used, why I should always double check listed specs to actual specs, and how to go about coding something like this.
7) Georgia Tech already disassembled the foosball table. It was just too big to keep around in precious senior design lab space. All the members of the team are graduated and on to the rest of our lives, so it's unlikely that any of us will work on this any time soon (not to mention, since we built it for a GT course, they own the IP rights to it). I'm hoping the professors will let students improve upon the design in the next couple semesters as part of their senior design.
DIY robotic foosball table is ready to throw down {Engadget}
Dec 26th 2007 11:36AM Hey, this is Shane Connelly (the team leader of this project). This article has been posted at a few websites and the same things seem to pop up time and time again (incorrectly).
1) That's actually me playing the robot in the video...and I'm actually pretty good at foosball. But it'd make for a pretty lame video if I played 100% against it ;)
2) The limiting factor was NOT the computer! Processing on a 800MHz Pentium 3 PC only introduced a total of about 1/100 second of lag -- most of the CPU cycles were spent asleep, waiting for the webcam to acquire the next image.
3) The main limiting factor was the mechanical build. None of us were mechanical engineers -- in fact, none of us had really done anything with gears/motors before this, either in class or out. Larger gears would allow the robot to move faster, but would have cost us $1500 -- wayyyyy out of our $500 total budget. Now that I've done it, though, I realize we should have just used a belt-driven system.
4) The second most limiting factor was the webcam. Philips sucks. Never buy a Philips webcam. The webcam we purchased, the SPC-900NC is listed as a USB 2.0 webcam with a frame rate of 90FPS @ 640 x 480. I think this must all be a windows driver trick where it just sends 3 frames for every 1 that it actually acquires, because the camera hardware is stuck on USB 1.1, with a maximum frame rate of 15FPS @ 320x240, in compressed image mode.
5) Yes, we realize the table wouldn't beat anything but the most novice players (not included in the video -- I actually programmed in insults that it shouts out the computer speakers whenever it scores). We built it under a "prototype" classification because we only had $500 and
