Subject:
|
Re: RCX & RIS, a fading glory?
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Thu, 30 Jan 2003 02:17:07 GMT
|
Original-From:
|
Steve Baker <SJBAKER1@AIRMAILstopspammers.NET>
|
Viewed:
|
1671 times
|
| |
|
|
Eric Sophie wrote:
> In lugnet.robotics, John Barnes writes:
> I was thinking that if a RIS 3.0 was made, I wonder if we have out grown the
> RCX? I know the list of things one could quote or list. But frankly, I have
> only seen a couple of hand fulls of Lego Robots that really use the RCX to
> its fullest. Dont's get me wrong. I say that statement very carefully.
> If one were to look at the range of RCX Creations. I would say it is far
> more difficult to make a Robot or device that really uses the RCX to its
> fullest,
I strongly disagree.
You can (with a lot of stupid messing around) push the RCX beyond anything
reasonable. The awesome Lego Typewriter is a classic example. These
are incredibly ingenious devices - and if you want to treat robot design
like solving a crossword puzzle, then that's fine.
However, some of us would actually like to design reasonably sensible
robots that do useful things rather than just being something to point
at and say "Wow! That guy has *way* too much time on his hands!".
Then, the equation is simple:
1) It takes a minimum of two controllable outputs to move a robot
under reasonable control along a specific path or to a specific
location. Two degrees of freedom - two outputs. That's for sure.
2) Ergo all the other aspects of the robot MUST be controlled with
the single remaining output. That's a very serious limitation.
There are sneaky mechanical tricks to get more out of what you have,
and some people find it fun to look for those tricks. However, for
many of us, that's a horrible pain in the neck. I want to build a
robot that can drive around the room *AND* have a controllable
robot arm on it - with at least three degrees of freedom. One
RCX can't do that.
3) It takes a minimum of two sensors to measure how the robot is
moving. Again, you have to measure your location in a 2D world,
you need two numbers to describe that location - and hence two
sensors...minimum.
4) Ergo, all the other aspects of the robot's perception of the world
MUST be funneled through the one remaining input. Again, there are
some dirty tricks that kinda help that - but still - if I want to
know whether the robot bumped into something - and whether it's
in the light or dark and what temperature it is and....whatever.
One RCX can't do that either.
5) Whilst the RCX has enough RAM for many applications - it would be
hard to fit a large neural network - or the progress of some evolutionary
learning algorithm - or even a simple map of the area it's explored.
6) Whilst the CPU is fast enough for simple things - it's way too slow
for image processing or anything like that.
7) Only being able to communicate back to base via a dog-slow IR link
is a serious restriction...it relies on having a line-of-sight and
it's *far* behind the state-of-the-art (IrDA for example). RCX needs
a radio link.
8) Debugging complex programs using only the LCD panel is tricky. Having
a pixel-addressable graphics display would be useful.
...I could go on.
All of our robots solve a small piece of the problem. There are things
like the LegWay - but it takes 100% of the RCX's resorces to just drive
the thing around...if you wanted it to fetch you a beer from the fridge,
you'd be screwed.
The trouble with all that is that 90% of the things I'd like to have in a
Lego robot controller are NOT needed in order to sell Lego to 12 year old
kids. Hence Spybotics.
However, what makes Lego's mechanical system interesting to little
kids, bigger kids and AFOL's is that it's open-ended and can be expanded
infinitely. A kid can enjoy a $5 set - a $20 set or lust after the $200
set - and an AFOL with *way* too much time and money can build an entire
city or a life-sized dinosaur with tens of thousands of dollars worth of
parts.
The computer system really needed to be designed on the same principles.
A modular computer - each part in a separate Lego-brick-sized package with
an inventive interconnection scheme and a plug-and-play scheme to enable
the CPU to figure out what's connected to it...this could support everything
from the very simple (1 CPU + 1 Pre-programmed-ROM + 1 Motor + 1 sound brick)
which could control an R2-D2...through to 10 CPU's plus camera plus 1Gb RAM
plus 40 motors and 50 sensors controlling an entire Lego train set.
Lego had the right idea when they came up with the infinitly expandable
mechanical system - they just didn't take the intellectual leap needed to
see that they exact same *IDEA* could be applied to the electronics and
the software.
> I say let me see the MOCs!
I can't build the things I want to create. I've tried - I have made
robots with 1 RCX and four Scouts to get me a dozen motors and sensors,
but then the comms system is too crappy, the enormous size of the
computer bricks forces me to build an enormous robot - which can't
drive fast enough because I can't get enough power to drive all the
motors I need from the non-modular batteries...you name it!
> I think that when the Star Wars movies conclude, as well as a couple of
> other Market lifespans slow down, we'll see some more new interesting things.
But that's not the trend.
We have Harry Potter Lego - Spiderman Lego...there will be more and
more movie tie-ins - because they are (presumably) successful.
As fast as one dies off, another appears.
However, I don't think those sets are the ones to worry about.
Aside from the god-awful micro-scout, the StarWars sets are actually
rather nice. The larger Harry Potter sets are basically just
castle stuff with a couple of specially themed MiniFigs.
The things I rail against are the Spybotics, Galidor, Bionicles,
Soccer and Basketball sets. These are useless for almost anything
other than building the model on the box lid...and that's NOT GOOD.
Spybotics *particularly* annoys me because it could so easily have
been exactly the thing we need.
---------------------------- Steve Baker -------------------------
HomeEmail: <sjbaker1@airmail.net> WorkEmail: <sjbaker@link.com>
HomePage : http://www.sjbaker.org
Projects : http://plib.sf.net http://tuxaqfh.sf.net
http://tuxkart.sf.net http://prettypoly.sf.net
|
|
Message has 1 Reply:
 | | Re: RCX & RIS, a fading glory?
|
| (...) Whoa, how about checking that attitude right there. Your inflection: "That guy has *way* too much time on his hands!" is a little offending. You speak of great concepts yet admit to not carrying them out. I build what I think I can accomplish (...) (22 years ago, 30-Jan-03, to lugnet.robotics)
|
Message is in Reply To:
| | Re: RCX & RIS, a fading glory?
|
| (...) I was thinking that if TLC links these two devices closely....well let's say I think that while they share simular characteristics they are really very different. I would hope that the fate of Mindstorms and Sybotics lay on different roads. (...) (22 years ago, 30-Jan-03, to lugnet.robotics)
|
33 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
|
|
|
|
