Subject:
|
Re: Vision on the Handy Board?
|
Newsgroups:
|
lugnet.robotics.handyboard
|
Date:
|
Wed, 10 Mar 1999 16:48:51 GMT
|
Original-From:
|
Paul E. Rybski <RYBSKI@CSstopspammers.UMN.EDU>
|
Viewed:
|
1230 times
|
| |
|
|
On Wed, 10 Mar 1999, Thomas Hauri wrote:
> we are using a NewtonLabs cognachrome system on one of our model cars.
> The car is equiped with a Handyboard, a cognachrom system, a camera,
> and a couple of RS232 to CAN converters.
I have a question for you guys. Why are you using the Handyboard to
control the car? The SBC332 board on the cognachrome system is a very
powerful microcontroller and is fully capable of handling the actuators on
your car. As an example, check out Newton Labs' page where they describe
"The Fast Vision Car" <http://www.newtonlabs.com>. I had a chance to
actually play with this thing in Portland at AAAI '96. The cognachome
system seemed to be the only micro on it and it drove that car extremely
well. Anne Wright (of Newton labs) said that when the cognachrome system
was first being built at MIT, they had it hooked up to a legged walking
robot which tracked light sources. The TPU on the 68332 is able to control
multiple servos and RC motor speed controllers (like the Blast! Duratrax
Mosfet motor driver). Newton labs supplies sample code for stand-alone
applications like this. One such code snippit is for a 2dof
servo-controlled camera head which tracks a bright orange fish swimming
around in a fishtank.
> Attaching a camera directly to the handyboard (using some sort of
> framegrabber) would ask too much of the 6811 I think althoough I have
> never tried it.
I haven't heard of anyone hooking a camera to a handyboard, but we did
have a guy in our lab build his own 6811-based microcontrollers and hook a
grayscale camera up to one of them. There's not much documentation left
from this project, but the URL is:
http://www.cs.umn.edu/research/airvl/walleye/
Walleye had three 6811-based microcontroller boards on it. One was the
frame grabber board which shared its 32K RAM chip with the camera. When
it wanted to grab a frame, the micro would go into a wait loop, running
code in either EEPROM or buffalo ROM (I can't remember now) until the
camera finished dumping the image into memory. Then, it would run
routines written in assembly (using the GNU gcc-based 6811 cross compiler)
to do simple object and shape recognition. It was fast enough to track a
black squiggle ball at around 10Hz, which was pretty cool. There was also
code written to recognize paper with the letter "T" and "O" on them. A
specialized high-current driver 6811 board was used for driving the motors
and the big gripper and a third was the "master" board which was used for
communications to the outside world and controlled the other two.
Communications was achieved through the SPI bus. A writeup was published
in the special vision issue of the Robotics Practitioner, Spring 1996. Vol
2, Number #2.
-Paul
---
Paul E. Rybski --- http://www.cs.umn.edu/~rybski
|
|
Message is in Reply To:
 | | Re: Vision on the Handy Board?
|
| Hi Bill we are using a NewtonLabs cognachrome system on one of our model cars. The car is equiped with a Handyboard, a cognachrom system, a camera, and a couple of RS232 to CAN converters. Our Handyboard has a CAN controller attached to it for (...) (25 years ago, 10-Mar-99, to lugnet.robotics.handyboard)
|
4 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
