To LUGNET HomepageTo LUGNET News HomepageTo LUGNET Guide Homepage
 Help on Searching
 
Post new message to lugnet.faqOpen lugnet.faq in your NNTP NewsreaderTo LUGNET News Traffic PageSign In (Members)
 FAQ / 551
550  |  552
Subject: 
[MindStorms FAQ 2.15 My robot runs great, looks great, and stays in one piece even when my kid throw
Newsgroups: 
lugnet.faq
Date: 
Tue, 13 Jul 1999 21:47:40 GMT
Viewed: 
1931 times
  
Subject:          2.15 My robot runs great, looks great, and stays in one piece
even when my kid throws it out the car window. But it's dumber than mud. What
can I do?
Content-Language: en
Topic-Level:      0
Revision:         Robert Munafo, 1999-07-13
Location:         /robotics/rcx/
Comment:          Topic-level is inaccurate pending a reference for topic
levels

<p>First of all, try not to get frustrated. You have encountered the same
issue robotics researchers have encountered for nearly fifty years.
Solutions to problems that seem simple to humans, even when you've
thought out the details of how it will work, totally fail to work when
put into practice. </p>

<p>It is actually just another case of the Fundamental Building
Challenge.  There are many different ways to approach the same problem
but only a few will actually work. In this case though, it's not just
a physical or aesthetic challenge but something abstract, probably
totally unlike anything you've ever thought about before. You have
limited choices and you don't even know what the choices are. </p>

<p>Many approaches have been suggested to deal with solving the abstract
challenges of robotics.  The MindStorms literature suggests you "play
computer" or imagine yourself physically in the position of the robot
as it does its simple actions, and see how it shuld react. But that
won't be enough to come up with a totally different approach to the
problem. </p>

<p>One important principle to remember right away is: </p>

    <blockquote>
    "Output" is much easier than "Input" </blockquote>

<p>In other words, it's much easier to make to robot <i>create</i> a complex
sequence of actions than to make it <i>recognize</i> something complex. </p>

<p>This generalizes to the basic principle of striving to simplify the
design, and simplify the problem you're trying to solve. </p>

<p>Some tasks might seem impossible. Suppose you want your robot to
follow you around like a puppy-dog. How are you going to accomplish
that? How can the robot distinguish <i>you</i> from everything else around
that <i>isn't</i> you? You probably can't -- even if you are the only
moving object in the robot's environment. </p>

<p>But, if you're willing to wear an infrared LED that continually blinks
a repeating pattern, the challenge suddenly becomes much easier. Using
NQC or Visual Basic, monitor the light sensor to detect the pattern
and check the relative level to gauge whether your direction of
movement is getting "warmer" or "colder". (By the way, a repeating IR
beacon is easy to make -- just take a TV remote control and tape down
one of the buttons.) </p>

<p>The best tool for simplifying designs and finding new solutions is to
go out and see how other people have approached their problems.  Spend
a while going around the MindStorms web ring and visiting other
robotics sites. One person's line follower solution might give you
just the idea you need to perfect your M&M candy sorter. </p>



1 Message in This Thread:

Entire Thread on One Page:
Nested:  All | Brief | Compact | Dots
Linear:  All | Brief | Compact
    

Custom Search

©2005 LUGNET. All rights reserved. - hosted by steinbruch.info GbR