Announcement

Collapse
No announcement yet.

Cold robot behavior

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

  • Cold robot behavior

    Last year at our state competition at Iowa State University, the temperature was around 10 degrees F. Unfortunately, our pit was several buildings away and we transported our robot to the competition building just prior to our first run. When the team started the robot it was programmed to go straight using gyro. Instead it veered immediately to the right. This had never happened before in practice or competition. The team reset and tried several more times and tried some other runs, but the same behavior occurred and they scored zero points.

    The team suspected the cold had affected the gyro and after we had been inside for about 30 minutes, the robot resumed normal behavior. We decided to not take the robot outside again and camped out in the competition building. The problem did not repeat for our later runs. One of the engineers serving as a judge was skeptical that the cold would affect the gyro and suspected a frozen large motor.

    We we plan to transport our robots in temperature controlled boxes this year, but I

  • #2
    Electronics are affected by cold, but 10 deg F for a short trip between buildings isn't cold. Given that you would move quickly while outside and arrive a little early I would suspect the robot was pretty close to room temperature by the time you started the first run. This problem was posted a year or two ago and I ran some tests during the middle of a MINNESOTA winter. On a sub zero day I left the robot outside for half an hour and then brought it in and ran some tests with the sensors and motors. I didn't see anything odd with the sensors. My motors were a little growly until they warmed up and my tires didn't have much traction due to condensation.

    If you want to test you can pop your robot in the freezer for a while, take it out and run some missions. The freezer isn't as cold as 10 deg F, but you can compensate by leaving it in longer and giving it less time to warm up. If you run this test I would be interested in your results.

    Using the gyro sensor to drive straight is a bad idea. To many things can go wrong that the tiny benefit of maybe going a little straighter isn't worth the risk.

    Comment


    • #3
      (edit to my original post by phone that cut off)

      Thanks Dean,

      I will run that cold test after our regional competition this weekend and report back to you. As to driving straight by gyro, I convinced my team to stop using it for shorter stretches, but they like it too much for distances over about 20 cm. The good news is that this year we used significantly more wall following and squaring on mission models. This has eliminated all but two "straight by gyro" sequences. Robot performance (at least in practice) has been much more reliable as a result.
      Last edited by jasponti; 12-04-2018, 09:26 AM.

      Comment


      • #4
        Okay, I've finally had a chance to run a cold test and I think I found the problem. First, I placed a robot in the freezer for 3 hours and then powered it up. First observation: the liquid crystal display on the brick was VERY slow (about a 2 second delay), but still worked.

        I had the robot run through a series of drive straight tests using drive tank, drive steering, and then drive with gyro assist. I then ran a series of basic turns and then some gyro turns. The robot pulled significantly to the right when trying to drive straight (but not as much as it did last year during competition). It was particularly noticeable when the robot first started moving and the robot was straighter a full speed. The gyro appeared unaffected by the cold and the odometry readings on the motors was almost identical after driving for 25, 50, and 100 cm.

        I flipped the robot to look for any cables that might have been causing problems and I think I found the problem. There was heavy frost on the metal caster wheels. The team's robot used two this year for added stability and turning the metal ball showed places of thick frost buildup. I used a microfiber cloth to clean the casters carefully and the robot was able to drive straight again.

        I worked very quickly because I didn't want thawing to big a big factor. Moral of the story (I think): watch out for ice buildup on the metal caster balls in cold weather!

        Comment


        • #5
          Originally posted by jasponti View Post
          Okay, I've finally had a chance to run a cold test and I think I found the problem. First, I placed a robot in the freezer for 3 hours and then powered it up. First observation: the liquid crystal display on the brick was VERY slow (about a 2 second delay), but still worked.

          I had the robot run through a series of drive straight tests using drive tank, drive steering, and then drive with gyro assist. I then ran a series of basic turns and then some gyro turns. The robot pulled significantly to the right when trying to drive straight (but not as much as it did last year during competition). It was particularly noticeable when the robot first started moving and the robot was straighter a full speed. The gyro appeared unaffected by the cold and the odometry readings on the motors was almost identical after driving for 25, 50, and 100 cm.

          I flipped the robot to look for any cables that might have been causing problems and I think I found the problem. There was heavy frost on the metal caster wheels. The team's robot used two this year for added stability and turning the metal ball showed places of thick frost buildup. I used a microfiber cloth to clean the casters carefully and the robot was able to drive straight again.

          I worked very quickly because I didn't want thawing to big a big factor. Moral of the story (I think): watch out for ice buildup on the metal caster balls in cold weather!
          When the robot crabs there's not much you can do about it. When I use a gyro and run my robot on carpet it drives at about a 15 degree diagonal. The robot is pointing in the right direction, it just doesn't drive in that direction.

          Comment

          Working...
          X