DPRG List  

[DPRG] Spring Robot Rules

Subject: [DPRG] Spring Robot Rules
From: Dean Hall dwhall256 at gmail.com
Date: Mon Feb 18 20:56:38 CST 2008

On Feb 18, 2008, at 18:57 , dpa wrote:

> These 4 challenges are basically navigation tasks in escalating
> difficulty: 1D (straight line), 2D (square), 2D with obstacles
> (#3), and 3D with obstacles (#4).

<joke type="nerdy" color="off" funny="little">
With Ackerman steering and bang-bang steering control, Argonaut will  
not drive a square for the second task, but more like the shape an  
inebriated unicyclist makes when attempting to drive a square.

> So although the challenges get more difficult, the task, i.e.,
> keeping track of one's location in increasingly difficult
> circumstances, remains the same.  And the metric of that task,
> the robot's offset from the origin, also remains the same.

So, for the first challenge (Drive from point A to point B and back  
to point A), if Argonaut's evil twin drives from point A to point C  
and back to point A, he might have a chance of winning even if point  
C is not near point B?

That's a rhetorical question showing the trouble of minutia when  
making rules.  I completely understand the intention to first get a  
group of solid competitors before worrying about rules.  That's part  
of the reason I made the first SHARP Dry Run an "event" rather than a  
contest.  I'm not suggesting you change anything by saying that.  I'm  
agreeing with your plan.

> Dean, how's the Argonaut robot and its new Kalman filter
> progressing?    Do you think it will have obstacle avoidance
> working by May's contest?

Argonaut's hall-effect encoders are working, PWM is working, the PI  
controller is working (the D had negligible effect), and I got GPS  
going and NMEA parsing working this past weekend.  Now I'm working on  
two servo device drivers for steering and sensor turret.  Next, I'll  
build out the body more and mount the ultrasonic and IR sensors.   
Then the I2C device driver for the compass.  *Then* the Kalman  
Filter.  Come May, I will have met many obstacles, but I have no idea  
how Argonaut will fare with those before him.

Things that haven't gone well so far: 1) the encoders don't work well  
at low speeds.  2) The RC car was made for speed, so it's motor is  
very close to stalling when driving at pedestrian speeds.  This means  
it has little torque yet consumes significant current if it has any  
resistance.  The motor driver reaches thermal shutdown when driving  
on a lawn.  3) The arm-gcc tools are not as easy to learn as avr-gcc  
and the CMUcam3 is hard to build on a Mac where the intended  
toolchain isn't available.  (I've aborted using the CMUcam3 for this  
Spring, I'm onto plan B)

Things that have gone better than expected: 1) The motor driver shut  
itself down rather than going up in smoke (tip o' the hat to the guys  
at Pololu).  2) The GPS module hack is working well and NMEA data is  
coming in nicely.  3) The AvrX RTOS works great and gives me a lot of  
flexibility in how I architect the code (thanks Larry B.!).  4) My  
home-designed ATmega103 board is still working after 8 years of abuse  
and neglect.


More information about the DPRG mailing list