Subject:
|
Re: Brainstorms
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Fri, 9 Aug 2002 03:08:07 GMT
|
Original-From:
|
Steve Baker <sjbaker1@airmail!antispam!.net>
|
Reply-To:
|
sjbaker1@airmail.NOSPAMnet
|
Viewed:
|
828 times
|
| |
|
|
Ralph Hempel wrote:
> 1. The RCX by itself is VERY powerful. It has approximately the same
> memory footprint and substantially more power than the on-board computer
> on the Apollo 11 spacecraft. What it lacks is I/O. It does not need
> a floating point processor, 50 Mhz clock, 512K of RAM, LCD touchscreen,
> or voice synthesizer.
It doesn't need a voice synthesizer - it DOES need the ability to ADD a
voice synthesiser.
It's a pretty pathetic CPU by modern standards...if all you are using it
for is turning motors on and off and reading sensors - then it's fine - but
once you can attach a camera and you want to do image processing then it's
pathetically slow...and I *want* to attach a camera to it. Then there's
neural networks, speech processing...all sorts of things a robot could
benefit from doing.
So what there is now is good enough for the peripherals we have now - but
if you want more peripherals then pretty soon your CPU will need to catch up.
> 2. The standard firmware is very good for its stated purpose, which is to
> introduce robotics to a population that is ill-equipped to deal with
> the actual complexities of real-time operating systems.
Yes - and they had the foresight to let you dump it out if you want to
run LegOS or something.
> 3. Add-in firmware (like my pbForth) removes many of the limitations
> of the standard firmware at the expense of complexity.
Yes - but if you want the same computer to work like a Scout (for people
who don't want to program) *and* like an RCX *and* like a Spybot - then
you might want to have a replaceable ROM in the package.
> 4. If we add a simple I/O expansion in combination with some additional
> firmware we could increase the capabilities of the standard RCX. A
> single sensor port could be used as a high-speed serial interface
> with this custom firmware. The advantage is that the adder modules
> could probably get their power from a standard bus and comms from
> the RCX. I already have an additional driver that lets you control
> up to two servos from one motor port.
Yep - but the extra electronics it takes to decode your serial port and drive
(say) a motor ends up considerably larger than an entire microprocessor.
If you are going to all that trouble, you might as well replace the whole
thing.
> 5. Most of the likely purchasers of RCX expansions are Universities
> or Colleges. They would also be looking at Basic Stamps or Handy Boards.
> They probably already have RCXs. Why not leverage this and design a simple
> adder for the RCX instead of a whole new device.
Because I think you can do much better - and stay within the SPIRIT of
Lego mechanical parts by making it (almost) infinitely expandable.
> 6. Designing a whole new device means providing support in the form of
> a compiler, run-time library for peripherals, probably an IDE, and
> of course a website and community for exchanging messages. This is WAY
> more work than anyone thinks it is.
No - it certainly isn't. You can make the software understand the RCX's
bytecode - so at a minimum, all the existing tools 'just work' (NQC, the
Lego environment, and all the other byte-code-based languages). Legos
would be toast though. However, it might not be that hard to port it
(I bet it's mostly written in C) - and we *DO* have compilers, IDE's and
all the other goodies in the shape of the GNU tools. The GCC compiler
has backends for so many processors that I've completely lost count.
I don't think software comes close to being the biggest problem...but
then I'm a full time programmer - and these things just don't seem that
hard to me.
> I'm rambling, and could go on and on. The biggest thing to understand is that
> the vast number of hobbyists never get beyond a simple bumper car with their
> RCX. The reason is that computer science, mechanics, and the required math are
> very difficult subjects to understand. You can't put a GUI around it and make
> it simple.
Yes - that's true...but Lego sold a *bunch* of these things and made good money
at doing it.
> As I'm sure I've mentioned before, I teach embedded systems design seminars
> and am generally dismayed at the level of understanding that most programmers
> have of these systems. We need a robust way of adding IO to the RCX, not a
> super-RCX.
But the RCX is little more than a single-chip micro - with a bunch of I/O. When
you've replaced the I/O, you've replaced EVERYTHING except one chip - the CPU.
I'd maintain that it's conceptually simpler to dump the CPU too and remove any
nasty restrictions that the RCX is going to impose on you.
> Maybe I should rephrase that. *I* need a robust way of adding I/O. I don't
> see a big enough market for a super RCX that would make it worth anyone's
> time.
Sure - as a DIY project that you don't expect anyone else to have to replicate,
it's easy...but expecting other people to cut holes in their RCX's case and
solder new wires all over the place is much harder.
The problem with all of this enthusiasm is that unless Lego do it, it's not
going to be a huge success - and we *know* what Lego have decided to do...
and many of us don't like it. Quite what the answer is to that - I don't know.
----------------------------- Steve Baker -------------------------------
Mail : <sjbaker1@airmail.net> WorkMail: <sjbaker@link.com>
URLs : http://www.sjbaker.org
http://plib.sf.net http://tuxaqfh.sf.net http://tuxkart.sf.net
http://prettypoly.sf.net http://freeglut.sf.net
http://toobular.sf.net http://lodestone.sf.net
|
|
Message is in Reply To:
 | | RE: Brainstorms
|
| (...) <SNIP> The more things change..... It's funny, but if you look back at the rec.toys.lego archives around 1998 you will find a thread that is very similar to this, and then LEGO brought out the RCX. Producing, selling, and supporting single (...) (22 years ago, 9-Aug-02, to lugnet.robotics)
|
53 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
