Announcement

Collapse
No announcement yet.

Is playing "defense" a legitimate strategy?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    I can also vouch for the value of anti-autonomous programs. As we have not created a reliable scoring autonomous yet, we have been blocking most matches, with a nearly 100% success rate. We have also used the same code to block those trying to block our partner's autonomous. As the Sparbots said, it is not a crowd pleaser, but it is a 50 point swing.

    3. Build your drivetrain with enough traction/mass to simply shove teams out of the way.
    I agree with Skinkworks: why is this risky? True, you are usually less agile when build this way, but you are also fairly unstopable. The main risk is burning motors, but a concientous driver should be able to know when to stop.

    I wonder whether teams have tried using cameras and image processing algorithms to navigate during autonomous.
    Last year we had an algorithm designed to find our position based off of the two corner IR beacons. We dropped it simply because the IR Seeker does not work at those ranges (12+ feet), and we were unable to build a custom IR sensor with a range better than the IR Seeker. Cameras would be quite interesting, but would probably require more processing power than the NXT can provide. Arduino can probably run off of the SuperPro, but that also seems too slow. What sort of processing power would this require, anyway? Perhaps something to play with over the offseason... ;-)
    FTC Team 4508
    E=MC Squirreled
    Website: emcsquirreled.github.com
    Github Repo: www.github.com/emcsquirreled/FTC-2012
    Lead Programmer and Electronics

    Comment


    • #17
      I'm not sure that an NXT couldn't manage a camera; suppose you use a simple 320 x 240 QVGA camera (used to be a standard for the resolution of video frames); that's a measly 76,800 pixels. I imagine RobotC could iterate through an image of that size without too much trouble. Or, if that's too much for the NXT's 4 MHz processor, I suppose you could build a small array of photodiodes.

      An Arduino is legal, but I don't think the NXT could supply enough power through the RJ12 cables for it to run properly; maybe it would work if the SuperPro could be wired to the battery pack, but I can't see a way for small onboard computers to be practical otherwise.

      Comment


      • #18
        Originally posted by Skinkworks View Post
        Another option in your list would be to quickly write an autonomous to counter your opponent's anti-autonomous: an anti-anti-autonomous. Useful in semirare cases (My team has done it before).




        Why is this risky? This is basically our strategy this year, and it seems to be working really well. We have one motor per wheel geared down 3:2 and the robot weighs 30 pounds. Very few robots can push us.
        Risky in the sense that there's always a bigger fish. The other team from our town weighs close to 40 pounds for example, and your wheels really only need to slip once to start messing with the program.

        Comment


        • #19
          Originally posted by Sparbots View Post
          I'm not sure that an NXT couldn't manage a camera; suppose you use a simple 320 x 240 QVGA camera (used to be a standard for the resolution of video frames); that's a measly 76,800 pixels. I imagine RobotC could iterate through an image of that size without too much trouble. Or, if that's too much for the NXT's 4 MHz processor, I suppose you could build a small array of photodiodes.

          An Arduino is legal, but I don't think the NXT could supply enough power through the RJ12 cables for it to run properly; maybe it would work if the SuperPro could be wired to the battery pack, but I can't see a way for small onboard computers to be practical otherwise.
          The Landroids custom-built a set of photodiodes (4x37) for the purpose of navigation last year. They mention it in passing in their autonomous videos: http://www.youtube.com/watch?v=Q6BVQ4iWWFY

          Comment


          • #20
            The Landroids custom-built a set of photodiodes (4x37) for the purpose of navigation last year. They mention it in passing in their autonomous videos: http://www.youtube.com/watch?v=Q6BVQ4iWWFY
            Very impressive! I guess it can be done after all. Again, another project for the summer when there is more time.

            An Arduino is legal, but I don't think the NXT could supply enough power through the RJ12 cables for it to run properly; maybe it would work if the SuperPro could be wired to the battery pack, but I can't see a way for small onboard computers to be practical otherwise.
            It depends on which Arduino is used. The SuperPro provides either [email protected] or [email protected] Either mode should work, depending on the model chosen. Arduino is (I believe, I could be wrong) 8 MHz, so you should be able to do moderately well on simpler tasks. No HD facial recognition, but a low-res navigation system just might work.
            FTC Team 4508
            E=MC Squirreled
            Website: emcsquirreled.github.com
            Github Repo: www.github.com/emcsquirreled/FTC-2012
            Lead Programmer and Electronics

            Comment


            • #21
              Originally posted by 3608robotics View Post
              I would be interested in seeing a robot bypass an anti-autonomous. There are a couple of main ways to do it that I can think of:
              1. Have an arm that can reach over other robots.


              1) can be done (off the top of my head) only by a diagonal scissor lift.
              Our team is a rookie Girl Scout team and they don't have any experience with mechanical assemblies at all. So over the Christmas break I built a second robot to demonstrate how a gear reduction, a pan/tilt mechanism, and a screw lift worked. The second purpose was to provide a blocking robot for them to practice driving around. They incorporated a screw lift but decided to keep their own design. They were having a great deal of problems with jerky turns so they decided to gear down the drive wheels. So the exercise was a success and the girls learned a lot.
              I had measured the distance from the dispenser to the near post and figured an arm could cover that distance and could reach over any intervening robots. It also blocked the route around that side of the rack. I had not anticipated that a team could just ram us and push us out of the way. Having seen on YouTube how few rings were actually being placed on the rack during the match, I decided to forgo the top rack and anything but the near post. It turned out that a single arm driven by as massive reduction gear box could just reach from the dispenser to the lower and middle pegs. The arm was built like an excavator that folded up inside the center of the robot. It was possible to place eight or more rings on the pegs and that seemed to exceed most of what was being successfully placed in the matches I had seen. The girls also saw how it wasn't necessary to attempt ALL the tasks in the competition, but to focus on a few and do those well. They didn't like the idea of a stationary robot so they kept their mobile design. They also decided to focus on placing more rings on lower pegs than build a more complicated design that would get them to the top row. It was cool to see them work together to analyze the pros and cons of their design and stick to their guns. (Figuratively of course.) It was also nice to see them reject the concept of a purely defensive robot even though they were struggling to successfully place any rings at the time.
              (Sorry to ramble on.)
              Neil
              _____________________________________
              Co-Coach - Team 6111
              Houston, Texas

              Comment

              Working...
              X