Subject:
|
Re: New contest
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Fri, 9 Dec 2005 20:46:29 GMT
|
Viewed:
|
5379 times
|
| |
|
|
I take it you are thinking about using the CMUCam or some similar
system with some level of intelligence on board the camera processor.
After all -you are really not going to be able to shift the amount of
data required to process on board the RCX, and I very much doubt the
RCX has anywhere near the amount of power requried to do any image
processing on board. I am sure something like striplight direction may
be acheivable with a very simple low-resoltuion CCD though.
In fact, a 4 x 4 grid of photosensors could give you orientation,
though not position info. Couple that with the I2C interface, and a
good striplight - and you have a basic starting point.
Taking it furthar - you could go for a triangle of three bright point
lights, and try those with the 4x4 grid - you may be able to infer
position as well as orientation if you go for an irregular triangle
shape.
With a couple of IR transmitter beacons - that is something the Spybot
brick is maybe quite suitable for. I have considered coupling my
spybot with the RCX to use its position sensing IR.
--
Danny Staple MBCS
OrionRobots
http://orionrobots.co.uk
(Full contact details available through website)
On 08/12/05, Mr S <lego-robotics@crynwr.com> wrote:
> Geoffrey,
> It is not my own plan of attack, so to speak, but it
> seems to me that there might be advantages to a
> florescent strip as opposed to a single bulb type
> light. Well, there might be if your camera system can
> differentiate alignment of the strip. That is to say
> that if your light strip runs due east/west, then you
> have only to look at the light and have a 50% chance
> of knowing exactly which direction east is. That means
> that your other landmarks are instantly easier to
> use/identify, or might be.
>
> Its all down to how your camera system works. I don't
> think there would be too much advantage to it if you
> are simply using light detectors with low resolution.
> Anything much lower than a camera would seem too low
> to detect the direction of the strip (but I could be
> wrong). My first thoughts would be to use a camera and
> the I2C interface for RCX from mindsensors.com...
>
> Once you can identify lights vs. bright spots on the
> ceiling, I wouldn't think the programming has to be
> more complex than that.
>
> Cheers
>
> --- Geoffrey Hyde <g.hyde@bigpond.net.au> wrote:
>
> >
> > "PeterBalch" <lego-robotics@crynwr.com> wrote in
> > message
> > news:200512070728_MC3-1-B1D9-C836@compuserve.com...
> > >
> > > Message text written by Mr S
> > >
> > > I'm not sure this rule is neccessary but it does • little harm. I expect
> > > that
> > > most of the "Don't get stuck" algorithms won't • need tremendous computing
> > > power. But if someone solves the coms problems and • comes up with a
> > > wonderful algorithm that involves taking bearings • off the ceiling lights
> > > and building a 3D model of the room then well
> > done.
> >
> > Just how much programming work would it be for a
> > robot's AI/inputs to have
> > to cope with flourescent lights (which basically are
> > a stripe and not a
> > point) on the ceiling??
> >
> >
> > Cheers ...
> >
> > Geoffrey Hyde
>
>
|
|
Message is in Reply To:
 | | Re: New contest
|
| Geoffrey, It is not my own plan of attack, so to speak, but it seems to me that there might be advantages to a florescent strip as opposed to a single bulb type light. Well, there might be if your camera system can differentiate alignment of the (...) (21 years ago, 8-Dec-05, to lugnet.robotics)
|
6 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
