Subject:
|
Re: rotation sensor mystery solved?
|
Newsgroups:
|
lugnet.robotics.rcx.legos
|
Date:
|
Wed, 26 Jan 2000 07:17:38 GMT
|
Viewed:
|
1184 times
|
| |
|
|
I put in some debug code to catch rotations which skip a state (transitions
like 3->1 or 0->2) which the legOS table `diff2change' currently translates to
0 steps (because we can't be sure whether that was 2 steps forward or 2 steps
back). These seem to be what are causing the last of my pendulum glitches.
To catch the intermediate states, I tried setting CKS in ADCSR (legOS calls
this ADCSR_TIME_134). This results in... pretty much nothing good. So
sampling faster is out, unless there's more to do when setting CKS.
Next I'm going to try adding momentum, and just assume that transitions which
move 2 states moved 2 states in the direction of the last motion.
Is anyone else working on this? Are people at least enjoying the lively
conversation I'm having with myself on lugnet.robotics.rcx.legos? If you want
to play along at home, take a rotation sensor, put 2 1x4 beams along one edge
of the top, put an 8t gear on a beam through the sensor and a 40t gear on the
center hole of the upper beam. Attach a stick of some kind with a weight (I'm
using an axle with a big tire) and swing it back and forth. You get a good 0
reference (straight down) so you can easily track when the zero point drifts
due to missed turns.
--Ben
|
|
Message has 3 Replies:
 | | Re: rotation sensor mystery solved?
|
| (...) Hi Ben, i think that's very interesting work you're doing, especially because i'd like to build sophisticated balancing bots, and already got 2 rotation sensors from lego world shop for this reason. I'm currently not actively working on that, (...) (24 years ago, 26-Jan-00, to lugnet.robotics.rcx.legos)
| | | Re: rotation sensor mystery solved?
|
| (...) It is very very interesting. I have worked the last weeks evenings to discover the fault in my concept, my modell and my Software. and I have discovered, that the fault can't be on my side. I discovered that my implementation of a fake (...) (24 years ago, 26-Jan-00, to lugnet.robotics.rcx.legos)
| | | Re: rotation sensor mystery solved?
|
| (...) I tried hacking something like this into the old rotation sensor code, and it didn't go real smoothly, but is definitely a necessity (I think) if you want the sensor to handle very, very high speed apps. (...) I am. I'm almost tempted to steal (...) (24 years ago, 26-Jan-00, to lugnet.robotics.rcx.legos)
|
Message is in Reply To:
12 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
|
|
|
|
