|
[DPRG] Processing unit for mobile robot
Subject: [DPRG] Processing unit for mobile robot
From: Chris Jang
christopher.jang at yahoo.com
Date: Sat Apr 19 15:17:40 CDT 2008
> > Also a very interesting thread on the difficulties
> of using a
> > PC platform for robotics by Robert Scheer over on
> the SRSList,
>
> Answering this email, listing some of the robots in
> our club, it
> came to me to wornder: is there was even one example
> of a robot
> based on PC motherboard and Big Iron OS, that didn't
> have to
> have supporting microprocessors under it?
>
> As I tried to think of examples, I couldn't remember
> even a
> single robot that fit the description of PC based
> and no micros
> doing the real work.
>
> Can you or others point to an example? Do they even
> exist as
> real operational robots?
The winner of last year's SRS Robomagellan used a
Pentium-M mini-ITX mainboard on a RC truck chassis.
Signal processing was done in MATLAB. Here are some
photos of it.
http://golem5.org/robots/robothon2007/img4145.jpg
http://golem5.org/robots/robothon2007/img4177.jpg
http://golem5.org/robots/robothon2007/img4195.jpg
Here is a web page found with a google search "mark
curry intrepid".
http://jandmworks.com/intrepid.html
I believe that the robot, Intrepid, did not have wheel
encoders. So it lacked accurate odometry. The contest
course ran under a metal roof which distorted GPS
reception, causing robots to run off into the grass.
Despite this, Intrepid was very robust. It never
became stuck. It too drove off the path, realized it
had done so, and then executed some search pattern
behavior to find it's way again. When it finally saw
the traffic cone waypoint and drove up to touch it,
everyone was clapping. It was a most impressive
demonstration.
jBot most likely would have found the waypoint much
faster. It has the best mechanical design of any
amateur robot I have ever seen. The IMU and accurate
odometry could compensate for the GPS distortion.
I was reading an article yesterday about the
sacrifices organisms make to have proportionately
larger brains. A big brain is very expensive but
allows solving problems that are impossible with a
small brain. So it is a minimax rationale - there are
some problems that can not be solved without a larger
brain and those are the ones that convey a survival
advantage to greater processing power. I see a similar
logic with robot computers.
My experience is that a PC mainboard greatly
complicates a small amateur robot. If a robot design
needs a microcontroller only, then it will be much
simpler, more robust, smaller, have greater endurance,
etc.
I think though that with highly miniaturized
ultraportable PCs becoming mass market items (e.g.
Asus Eee PC), the design tradeoffs are shifting. If
you can buy a 1 GHz class palmtop that uses flash
memory for under $500, then it becomes cheap enough to
use for a robot.
____________________________________________________________________________________
Be a better friend, newshound, and
know-it-all with Yahoo! Mobile. Try it now. http://mobile.yahoo.com/;_ylt=Ahu06i62sR8HDtDypao8Wcj9tAcJ
More information about the DPRG mailing list
|
