332Re: [intellibrain] Re: Getting intellibot to work with Python
- Jan 6, 2009I think the intellibrain robot is great and you can to quite a bit with it -- much more than most other small robots. (I'm another user like you.)The issue is that if you want to do "real" computer vision, you need a lot more computing horsepower than you will find on the small CPU of small robots. If you want to use OpenCV, you either have to have a much larger robot that can carry around a full PC / laptop / linux server, or you need to have the vision processing on a wirelessly-connected external computer not mounted on the robot.However if you are allowed to plant color markers in the environment, the CMUcam2 is great because it can localize the colored markers. True computer vision is a very difficult and mostly unsolved problem. The CMUcam2 is a nice simplification as long as you have some control over the environment.Paul----- Original Message -----From: seanjm8pck18Sent: Tuesday, January 06, 2009 10:18 AMSubject: [intellibrain] Re: Getting intellibot to work with Python
I'm actually using the CMUcam2. The intellibrainbot was supposed to be
my project, but maybe it's not the best robot for this.
The robor is supposed to get from point A to B , based on self
localization and the pictures/video it takes. I wanted to use edge
detection... but I come to realize that I can't do a whole lot with
this robot...What can I do with this robot?
--- In intellibrain@ yahoogroups. com, "Paul King" <email@...> wrote:
> I have been experimenting with pyrorobotics (www.pyrorobotics. org),
a python robots toolkit that has a driver for the Intellibrain.
Python runs on the host and communicated to the intellibrain over a
serial cable or bluetooth interface. I have also been working on an
updated driver for this toolkit.
> Regarding using OpenCV, the Intellibrain as-is does not support
traditional computer vision and seems unlikely to work with the OpenCV
library unless you are mounting your own camera (not CMUCam) and using
either a USB camera interface (liek a webcam) to the host computer or
using a video capture card with a mounted camera of some sort.
> In particular, the CMUCam does not support full-scale computer
vision, that is, it is not designed to provide a stream of image
frames for downstream vision processing. Rather the CMUCam does very
simple on-board computer vision for the sole task of tracking color
blobs. This gives you a lot more information than an IR sensor, but
it is not full computer vision. You didn't mention CMUCam, so perhaps
this is not what you are using, but this has been a point of confusion
> ----- Original Message -----
> From: seanjm8pck18
> To: intellibrain@ yahoogroups. com
> Sent: Sunday, January 04, 2009 7:13 AM
> Subject: [intellibrain] Getting intellibot to work with Python
> I want to try and use OpenCV with the bot. But I also want to use a
> python ide to interface with the robot. Any tips? also, is it possible
> to just extract one frame and use it as a picture, and the process it?
- << Previous post in topic