Tom Handley wrote:
>   Dan, the `mechanical slop' is gear "backlash" and steering linkage. As
>someone else pointed out, you really need a platform with two motors to do
>what you want. R/C `racer' cars are really not suitable for real robotic
>experiments. Again, all I can do is go forward and backwards, and based on
>mechanical and IR sensors, steer one way or the other while navigating.
>Again, I was bored one day and hacked up this car...


Hi Tom,

Finally found a useful RC-car hack. They used a Nikko Hummer, Ming
xmtr/rcvrs, and AVR controllers. They addressed some of the issues you
mentioned, like steering, and motor control. USed a servo for better
steering control, and designed an H-bridge for motor control that
they say produces snail speed up to original - I imagine not much
torque on the low end:

http://instruct1.cit.cornell.edu/courses/ee476/FinalProjects/s2000/ameer/

From their comments, it sounds like you may be right about problems
inherent with using an RC-car for robotic purposes, still some useful
info on the site.
================


>   So far my PID is fairly crude. It does P and I but not D... Given that
>this thing is suppose to wander around the house and it's around 14" long,
>it tends to rely on bumper switches more than the IR to get out of trouble
>;-). I've reached the point, and have the level of interest, that I want to
>go forward with a new platform with two motors for steering. Again, as
>someone mentioned, it would be nice to rotate around a point.
>

The latest Hummers by Nikko tout hi/low speeds and turbo burst
- not sure how they get this - certainly not gearing - maybe 2-speed
motors [this would be nice] - maybe simple current control [probably
marginally useful].

But still, according to the site cited above, the front axle of the
Hummer was pretty flimsy and they replaced a solenoid with a servo
for better control. Sounds non-optimal, but a possible.
====================

>
>   I too have the same goals as you as far as a platform for adding sensors
>and studying control methods (ie: Fuzzy Logic and others). I realized I
>would not be able to do that with this platform. Still, it has allowed me to
>study things I've long been interested in such as H-Bridges, PID, and
>sensors related to an autonomous platfrom which I can apply later.
>

Even with a non-optimal platform, sounds like you have a good start
on your project. Seems like you could think of it as a project where
higher-level control is used to help compensate for lower-level
weaknesses --> I think you'll find those Rodney Brooks' papers I
cited last time concentrating on systems designed precisely to do
just that.

And of course, fuzzy logic controllers are designed to do this too.
=====================


>   Ya know... Given the talent in this group, I wonder if we could form a
>company to bring simple robotics down to a reasonable price-level?
>Parallax's pricing is ridiculous... As most have noted, buying a simple
>platform with gears and motors is ridiculous... Anyway, just a thought...
>

Ahhh, grim realities of the market place. Remember the CUMP - original
idea was a $10 item that could replace multiple items costing lots
more. At any rate, most robot kits/etc I have run across so far do
not seem to be very expandible, or else are just the bare basics. I
suspect most people spend the majority of their time working from
the ground up, playing with the mechanical platform, getting to the
point you mentioned, and then go on vacation.

OTOH, I am most interested in getting past that point, and on to
more interesting higher control. If I actually tried to sell something
here, it would probably be geared towards adding higher-level
functionality to lower-level devices, ala Brooks.

best regards,
- Dan Michaels
==============

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu