Announcement

Collapse
No announcement yet.

line tracing with very large wheels

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

  • line tracing with very large wheels

    Our robot has been tracing line really well with the standard ev3 education wheels (56 mm diameter). Recently we have been testing with larger wheels (68.8 mm or 81.6 mm) and suddenly the line tracing (proportional line tracing) becomes very hard. We tried all kinds of parameters but the robot is either not able to get back to the line, or wobble to much, making it not traveling straight.

    Any suggestions?

    Thank you very much!

  • #2
    What kind of algorithm are you using? Posting the code might help us provide better suggestions.
    Kansas City Region Head Ref 2014-present
    KC Region coaches and teams can ask FLL robot game rules questions at [email protected]

    Comment


    • #3
      As a very rough rule of thumb, if you double your wheel size, keeping the same track, you need to divide your proportional gain by 2.

      If that doesn't work look at what else changed when you changed wheel size.

      Comment


      • #4
        Is the light sensor higher off the mat now? Bigger wheels also means the robot goes faster with the same power setting. You may need to slow it down.

        Comment


        • #5
          We were using standard proportional line tracing and tested all kinds of multiplier values. The robot is adjusted so that the color sensor is the same height above the ground as when we were using smaller wheels. The reason for us to try larger wheels is to make the robot faster. But when we set the power to make it move fast, the line tracing becomes unstable. Of course if the starting position is ideal, the tracing is ok. But we always test it by setting starting position a little off.

          Maybe PID tracing would work better.

          Comment


          • #6
            Originally posted by danioyuan View Post
            We were using standard proportional line tracing and tested all kinds of multiplier values. The robot is adjusted so that the color sensor is the same height above the ground as when we were using smaller wheels. The reason for us to try larger wheels is to make the robot faster. But when we set the power to make it move fast, the line tracing becomes unstable. Of course if the starting position is ideal, the tracing is ok. But we always test it by setting starting position a little off.

            Maybe PID tracing would work better.
            Unlikely. Integrator usually just makes the robot tracking worse. The best use I've seen for integrator is allowing a lower P gain while still tracking around corners. However this usually requires some unusual handling of the integrator sum to prevent your robot from turning the corner and continue turning. Tracking around corners is not a useful robot skill for most FLL challenges, while having a robot that doesn't spin in place while it is supposed to follow a line is most desirable.

            The light sensor spends so little time "seeing" the line once the robot starts to wobble that the error is constant (sensor all on or all off the line) and the derivative is zero. When the derivative is zero the derivative gain adds nothing.

            Comment


            • #7
              Agree that full PID in FLL-usage has never shown to be helpful in my basic experimentation, just an opinion. P is just fine.

              If the 'approach' to the line is unstable, but once-on-the-line is good...can the solution be broken up better into those two parts? One routine and P/Target/Gain values to "find" and ""orient" to the line, another to "follow", presumably at a higher Target lower Gain. Just like of FLL, don't digest the issue all in one gulp.

              Comment

              Working...
              X