Subject:
|
Re: monitoring motor revolutions
|
Newsgroups:
|
lugnet.robotics.rcx.nqc
|
Date:
|
Wed, 29 Nov 2000 13:44:12 GMT
|
Viewed:
|
1926 times
|
| |
|
|
In lugnet.robotics.rcx.nqc, Kris Zabriskie writes:
> I am currently working on a program that will monitor where a RCX goes. I
> am trying to establish a coordinate system so if it finds a light it can
> tell the location of the light. I was thinking about monitoring motor
> revolutions, but then I am worried about how to handle bouncing off of a
> wall. Does anyone have any ideas?
Hi Kris,
In addition to Lester's comment here some ideas and statements regarding your
project.
As Lester stated, you will need accurate distance, speed and direction
measurements - and that is so true. Furthermore, for calculating coordinates
you will need some math functions for sqareroot and most probably sine and
cosine, depending on what you are going to do with the robot. This can be done
but it is not quite trivial. However, the bigger problem is accuracy.
To get an understanding of this, you should definetly deal with some basics:
1. Make the robot drive a s-t-r-a-i-g-h-t line of about 2 meters.
2. Let it drive back exactly to the starting point.
3. Teach the robot to make a 360° turn. Then let it do a 90° turn 8 times in
succession.
This may sound trivial and stupid tasks - try them!
Well, I tried hard and here is the result.
Solving the above tasks is not possible, at least not accurate enough - even
when I used two rotation sensors. There is too much of backlash (better: play,
slip? - sorry for my bad English) in the gears. For about 60 cm the robot was
driving over a line but than it drifted away. Also the friction on the ground
that Lester stated is important (especially for the turn around job).
A solution would be putting a light ontop of your robot and a camera above your
working area. But then you need an image analysis program ...
Perhaps I am wrong but I think that the kind of tasks you are thinking of are a
little bit too much for the t-o-y robots. With line-following, wall-avoiding,
finding of light and objects, the limits of "move-around-tasks" are reached.
But as I said - perhaps my phantasy is too small and I may be too narrow-minded
due to my earlier job where I developed optical sensor systems and control
software for welding robots for years. In my thesis where I covered the whole
range of these tasks including image analysis I dealt only with robots that
were not driving around in a room - but that was hard enough.
It may sound that I am a bit frustrated about Mindstorms. But in fact this is
not the case. After doing some work on line-follwoing and associated tasks
(take a look at my model "Transportobot" at Mindstorms page) I am now
experimenting with two robots that interact in games, always ignoring my wife's
question "why can't you build something that is useful?". I would, if there
weren't these constraints ...
Hope that I didn't frustrate you and that my statements don't sound like
"listen, the big, wise teacher is telling you something" ;-)
Best regards
Bernd
|
|
Message is in Reply To:
 | | monitoring motor revolutions
|
| I am currently working on a program that will monitor where a RCX goes. I am trying to establish a coordinate system so if it finds a light it can tell the location of the light. I was thinking about monitoring motor revolutions, but then I am (...) (24 years ago, 28-Nov-00, to lugnet.robotics.rcx.nqc)
|
4 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
