Announcement

Collapse
No announcement yet.

REV 2m Distance Sensor

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

  • REV 2m Distance Sensor

    Is anyone having luck with the new REV 2m Distance Sensor? We did some very cool math to calculate our location based on sensors aimed at 2 walls... only to find that they need to be pretty close to perpendicular to the wall for an accurate reading.

    We dealt with that by putting the sensors on servos, but are finding that they do not give a true reading beyond about 4 feet. The REV website says up to 200cm - is this possible?

  • #2
    No, it is not possible. We used them all season, worked great. You need to consult the ST VL53L0X datasheet for the actual specs, the Rev verbiage is the max ideal condition spec. (the 200 cm spec is looking at a pure white wall, typical max spec on other surfaces is 80 cm.) We found that after about 60 cm on the plexiglass walls, the accuracy went south.

    We used it for measuring target heading vs actual wall angle. We used it for sampling distance at about 6 to 12 inches. There is a fair amount of jitter, ideally you should take three samples and take the mean, but the timing budget for that sensor is 33 ms and with the horrible FTC Java framework, this was not possible.

    Comment


    • #3
      11343_Mentor - Why don't you just write any code you need that is more timing sensitive with the Android NDK? You can write your own libraries there in C and C++, do your filtering there and then access it from your java?

      Comment


      • #4
        We were getting a maximum range of about 48 inches perpendicular to the walls, and at 24 inches from the walls we could still get readings up to +/- 30 degrees from perpendicular to the wall.
        We also saw the jitter in the readings of +/- 0.3 inches, but we could resolve that either by averaging, or because the PID algorithm that they fed had lag from the robot movement so could absorb the jitter.

        The two biggest issues we had were:
        1) Long read times of 20 to 50 mSecs, averaging at 35 mS. (Compare to Modern Robotics ultrasonic sensor which ranged from 10 to 30 mSecs, averaging at 15 mS)
        2) Zero offset - sensors had different zero ranges that varied by a couple of inches, and also seemed to change over time. We needed fairly accurate readings from two sensors to determine our angle to the wall so the poor zero offset threw this off. We handled this by having a calibration program that would take the sensor measured values and compare them to the know distance in a particular robot configuration, then write this to a file which was read back by the main autonomous program and used to fix the sensor offsets.

        We used the Rev 2m sensors because of the better range resolution than the Modern Robotics sensor, and we ended up with a pretty robust wall following method using them.

        Comment

        Working...
        X