Subject:
|
Re: Fishing for a tasty linkage/gearing/robotic building thread...
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Thu, 13 May 1999 13:09:24 GMT
|
Original-From:
|
Paul Speed <PSPEED@AUGUSTSCHELL.COMihatespam>
|
Viewed:
|
1597 times
|
| |
|
|
Mario Ferrari wrote:
>
> In lugnet.robotics, Anders Isaksson writes:
>
> > An idea we have (but no implementation ideas yet), is to drive the program in
> > discrete 'steps', sampling the sensors during the step, and decide the action.
> > This would make exact turns possible. The main problem here would be to
> > distinguish between the square blocks, and the round ones (try to measure 'on'
> > time?), and select different actions.
>
> I like a lot your idea of a "phisical" program stored on a 2 x n plate. You
> could use one sensor eg on the left side as a sort of "clock" mechanism which
> marks your steps. Doing so you very precisely know when to read the right
> sensor which is the actual code reader. If you use round bricks for the clock
> sensor you may stop the program movement when the clock sensor closes. On the
> right side you could simply put a 1x1 brick to mean '1' or nothing to mean '0'.
> To enhance the vocabulary of your "bytecode interpreter" all you have to do is
> to lengthen your "word" to the desired number of bits. Eg: reading three
> "bites" to have an 8 codes instruction set.
>
> > With light sensors instead, perhaps you can distinguish different colors? I
> > don't have any light sensors (CM, remember?), so it's just a guess.
>
> Yes with the light sensor you could use colors to select codes, but not more
> than three or four of them (in my experience).
>
> BTW I don't think this solution helps with the precise 90 degrees turn
> problem. I guess best solutions have already been suggested:
> - two indipendent groups of wheels one for the fwd/back motion and the other
> for left/right, with a third motor to lower/raise the proper group.
> - a rotation sensor to keep track of actual rotations.
>
> Mario
Or third, a synchro drive constrained to a ninety degree
throw. This is simpler than the independent wheels since it doesn't
require an extra motor or complicated mechanism for raising and
lowering. It is also easier than the rotation sensor version since
you may not necessarily have a rotation sensor.
The synchro drive has been discussed on here a few times
before, but I will reiterate some of it. I haven't attempted this
particular construction so I don't know if it can be done with the
stock Minstorms kit. (I also don't know what a stock Mindstorms
kit is anymore since all of my sets are now one big set.) :) I have,
however, experimented with various synchro drive... just not the
one I propose for your solution.
Imagine you have a robot platform with four wheels at
each corner.
Top view: (please use a non-proportional font to view)
+-----------------+
| W W |
| W W |
| W W |
| |
| |
| W W |
| W W |
| W W |
+-----------------+
Now, forward would drive the bot up the page and reverse would
drive the bot down the page. Each set of wheels turning rotation
is synchronized to some central source so they all turn together.
Top view:
+-----------------+
| W W |
| W W |
| W W |
| |
| |
| W W |
| W W |
| W W |
+-----------------+
Now the bot will drive at an angle. Note that the platform itself
never changes orientation. Continue the rotation to 90 degrees
and you get...
+-----------------+
| |
| WWW WWW |
| |
| |
| |
| |
| WWW WWW |
| |
+-----------------+
A robot that can drive left or right.
If you constrain the rotation to be only between 0 and 90 degrees
then by running your rotations extra time you can be sure that you
are either 0 or 90 depending on the direction you turned. Running
forward or reverse will then get you all four cardinal directions.
The most complicated part of this will be getting the drive train
snaked through whatever rotation assembly you have. Most of the
synchro drives you will see examples of use the turntable pieces.
I do not have any of these so I have experimented with several
other linkages. I won't go into them here... they are an
implementation detail.
The other hardest part is usually getting the rotation linked. With
turntables and some chains it is easy. Without either it is more
complicated if not impossible. Since in this example we only need
90 degrees rotation then it should be possible to link them together
with beams. This normally won't work because it limits the amount
of rotation. In this case that is a good thing.
I will experiment with this design a little bit this evening. I've
been looking for an excuse to play and I think this fits the bill.
-Paul (pspeed@progeeks.com, http://www.progeeks.com)
--
Did you check the web site first?: http://www.crynwr.com/lego-robotics
|
|
Message has 2 Replies:
Message is in Reply To:
24 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
|
|
|
|
