Joystick accelerometer

Hello all, I read in an old thread that the z-axis on the vexnet joystick doesn’t properly work. I’m wondering if this is still true, and if it does still not work then how did person(s) from vex control the armbot in this video: http://www.youtube.com/watch?v=HbxEerw6vk4

You don’t really need the Z axis if you just want to use it as for tilt control like they show in the video. I’ve used the X/Y accel data to tilt-drive a robot and it works fine.

You really only need 3-axis acceleration data if you are trying to detect actual 3-axis acceleration. But since the operator probably doesn’t want to have to flail around wildly to drive the robot, it would probably be of limited use for an actual robot control. You could perhaps detect the remote being dropped and treat that like an emergency stop :wink:

Cheers,

  • Dean

Alright, cool. Thanks

I assume you mean the section of video from 18 to 24 seconds where the forearm and wrist are following the movements of the controller.
I see the robot has two degrees of freedom in this section:

  1. forearm up-down
  2. write twist

Since the Joystick has two working accelerometers: X-axis and Y-axis
that should be enough for independent control of two degrees of freedom.

Note that the accelerometers are used as tilt-sensors in this context,
and the robot arm probably has potentiometers to report the twist of these joints. The code moves the motors controlling the joints to get the feedback to match the tilt of the controller. Or even better, mount an accelerometer on the hand and move the joints to get the two accelerometers to match.

Do you think it needs to be more complicated that that?

Another possible explanation is that the robot is doing a preprogrammed sequence, and the human has just practiced leading the same movements.

We were able to get a robot arm responding to human joystick movement as demonstrated in this video: http://www.youtube.com/watch?v=Y5tPkMS1_h8
(take a look at about 1:00). But there was really no way of fine tuning the responses. It was all programmed in extremes of movement, all in, or all out. The accelerometer was just too noisy to really finely position the arm.