Subject:
|
How can we make TINAH peck ? (long)
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Wed, 28 Aug 2002 22:16:18 GMT
|
Original-From:
|
Roul Sebastian John <roul_john@!spamcake!web.de>
|
Reply-To:
|
rjohn@*AvoidSpam*uos.de
|
Viewed:
|
736 times
|
| |
|
|
Dear LEGO experts,
we are currently running a research project using a LEGO robot and we
have some questions for which we would very much appreciate any help !
Let me first sketch the background: I am a Ph.D. student in the first
international Cognitive Science program in Germany, at the University of
Osnabrück (Nothern Germany). (In case you are interested in CogSci,
please have a look at: http://www.cogsci.uos.de/ and also at
http://www.eurocogsci03.uos.de/.) My focus is on trying to model animal
learning with robot models (an approach I called Comparative Cognitive
Robotics), and this interest now lead to a one year Master student
project called EROSAL (Empirical RObot Study of Animal Learning), which
aims at implementing a robot model of visual discrimination learning in
hens. More on this project you can find at our new homepage
http://www.cogsci.uos.de/~erosal/ . The LEGO Mindstorms robot we are
currently building we have named "TINAH", standing for "TINAH's Not A
Hen". But though she is not a hen, TINAH has to move and to peck. And
this is where our questions come in....
(1) TINAH has to peck at so-called pecking disks, which are small round
plates, onto which patterns are projected. (The task is to gather
information about these patterns and peck only at the "correct" ones.)
The height is not variable, so we can build a pecking mechanism at the
fixed height of the pecking disks. But the speed of the pecking should
be variable, i.e. it has somehow to be coupled to one of the RCX motors.
We have built a pecking mechanism that is based on an up-and-down
movement, but this mechanism is very brittle and most often makes some
important parts of the robot fall off (a phenomenon not often seen in
our model animals :-). So, what we rather look for is a mechanism which
turns the circular movement of an RCX motor into a back and forth (but
not up and down) movement in a stable way. Has anybody of you any
experiences or any suggestions how such a mechanism should be built with
LEGO ?
(2) It turned out that we will need more than one RCX to be able to use
more than those three sensors, and possibly also more than three motors.
In the robot model we have built until now, we combined two RCXs. But
unfortunately, there seems to be surprisingly little information on
using more than one RCX for building LEGO robots. In the many books on
Mindstorms I have, I only found some information on this in the last
chapter of the new Ferrari et al. book (about which I learned from this
list -- thanks a lot !). But even there, the RCXs have to be built into
two separate robots, separated by a wall, so this is also not applicable
in our case. What we would really need is a way to write to and read
from two RCXs which are controlling different parts of *one* robot.
Currently, our programming gurus are struggling with this and try to
define their own protocol etc.. But is there really nobody out there,
who already dealt with this problem and could help us, at least with
part of a solution ?
(3) In her environment, TINAH would have to move left and right. She
should best also be able to move forward and backward, but *without*
turning. Is there any mechanism implementable in LEGO which can move in
such a way ? (If not, we will leave out the backward and forward
movement, but this can lead to problems in the course of the behavioral
experiments we try to model.)
(4) Our vision experts wonder if there is a stable Linux driver for the
LEGO camera (which best should work with the Video for Linux standard)
or if we have to use a different camera. Does anybody of you have
experience with using the LEGO camera under Linux and could give us a
recommendation ?
We (i.e. the members of the EROSAL workgroup) thank you in advance for
any hints and tips on these issues,
greetings to all LEGO enthusiasts,
Roul Sebastian John
--
=====================================================================
Roul Sebastian John
=====================================================================
University of Osnabrück
Institute of Cognitive Science
Katharinenstr. 24
D-49069 Osnabrück
Germany
phone1: ++49 541 750 82 96
phone2: ++49 541 969-6225
fax: ++49 541 969-6210
=====================================================================
mailto://rjohn@uos.de
http://www.cl-ki.uni-osnabrueck.de/~roul
=====================================================================
Behaving like a robot usually does not make you build a better one.
=====================================================================
My brain -- it's my second favorite organ. (Woody Allen)
=====================================================================
|
|
Message has 2 Replies:
 | | Re: How can we make TINAH peck ? (long)
|
| (...) > So, what we rather look for is a mechanism which (...) It rather depends on how fast it has to go - and how much power it has to deliver (eg how heavy is TINAH's beak?). One way is to use a worm gear (one of those black spiral-shaped things) (...) (22 years ago, 29-Aug-02, to lugnet.robotics)
|
5 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
