DPRG
DPRG List  



[DPRG] Balancing robot difficulties

Subject: [DPRG] Balancing robot difficulties
From: Randy M. Dumse rmd at newmicros.com
Date: Mon Jul 2 14:51:18 CDT 2007

> And the video of the new robot is here: 
> http://www.allthatjazzagency.co.uk/robot/balance.swf

Sorry for my confusion.
	
>> Are you using quadrature? You mentioned "speed" of the motors
>> only.
>
> No, only speed - one encoder per motor.

I see. So you aren't really implementing a PD loop at all.
(Position, Derivative) but instead a Velocity loop. I think that
was David's approach as well.
	
> It's in the Khz range. I've forgotten ...

Okay, atleast it is above 40Hz, more appropraite for those
motors.

I think David's loop was low, in the Hz as well. I think SR04
was 32Hz. I don't remember exactly what nBot was. Think nBot was
a later processor than the HC11, but I'm not certain. jBot is
HC332 as I recall.

Do you have the option of easily modifying this loop timing? For
instance can you raise it by some odd fraction, to see if you
are accidentally hitting a mechanical resonance? Which might be
better or worse for only a slight change in loop repetition?

> The reason I'm not convinced about it is that
> it was SO easy to get the original, servo-based
> robot to balance, 

Well, your not giving the Servo manufacturers much credit for
their years and years of experience in Servos. They have dead
bands. They have carefully shaped response curves, and so on.
Their smooth operation didn't just happen. It (probably)
represents thousands of hours of engineering and
experimentation.

Watching the new video suggests the frame might be bending. I
suppose this by watching a gap between the bottom most bit (of
what looks like) electronics and the motors. It seems to change
width with the acceleration / decceleration of the base. If so,
this is inducing a pendulum like time constant that is screwing
with your angle reading, and would set up the oscillating
feedback. Again, it is hard to tell with the video, so it is
just a "longdistance" attempt at an observation.

Randy
www.newmicros.com


More information about the DPRG mailing list