Subject:
|
RE: Kalman Filter on the NXT
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Fri, 28 Dec 2012 23:09:26 GMT
|
Original-From:
|
Bruce Boyes <bboyes@systronix.!AvoidSpam!com>
|
Viewed:
|
20726 times
|
| |
|
|
I have a memory from attempting to use the Kalman filter, starting with a
model, and working through the math, that it made some assumptions, which
importantly included an originally good sensor signal subsequently clouded
by noise. Kalman can recover most of the original signal. We modeled this on
the PC and could see good efficacy in that circumstance.
But in our case we had a noisy sensor, and in this case my memory was that
the Kalman filter was not any solution. In our particular case we are trying
to use a (non-Lego) DC motor commutator as a tachometer, while the motor is
also being driven. We pause driving for long enough to capture some
commutations while the motor is coasting. We concluded that the quality of
the commutator was so low due to sloppy manufacturing that this corrupted
both the predictability of driving and sensing. Worse, this slop was not
consistent across many motors, so we could not make a universal model for
it. What I don't remember is how we concluded that there was a difference
between a clean signal later corrupted by added noise (both able to be
modeled), and a signal which had enough sensor variability (not simple
randomness) that it was essentially impossible to model.
These motors drove a tracked robot. This was a small chassis, under $100
(had we been able to spend more on the chassis I believe our problems would
not have existed, but then neither would our perceived market). To further
add to the problem, the molded rubber-like plastic tracks were also poorly
made. Some had spots at which they were harder to flex. We made some simple
fixtures using known good Lego motors as tachometers, with the motors
free-running (chassis not on a driving surface), with and without the treads
attached and could document the motor and track disparities. Under these
ideal circumstances we could not get a consistent tach signal from all
motors. We sorted through several tens of chassis and tracks. Perhaps 10 or
20 percent might have been usable. We finally gave up in despair.
If my understanding of Kalman capabilities is flawed I would love to hear it
since this idea might get resurrected.
Thanks, and I hope I did not make too much a fool of myself...
Bruce
|
|
Message has 1 Reply:
 | | Re: Kalman Filter on the NXT
|
| (...) That's what the Kalman filter is about. One of its important application areas is navigation (from rockets to ships and robots). In our example, we try to develop the implementation for a robot driving in one only dimension, the x-axis. And (...) (12 years ago, 29-Dec-12, to lugnet.robotics)
|
Message is in Reply To:
4 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
Active threads in Robotics
|
|
|
|
|
|
