Subject:
|
Re: Autonomous Robot
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Fri, 4 Aug 2000 00:03:42 GMT
|
Viewed:
|
837 times
|
| |
|
|
"Andy Gombos" <lego-robotics@crynwr.com> wrote in message
news:39888843.D78EFA17@yahoo.com...
> For mapping, you can use an ultrasonic sensor, like the one designed by • John
> Barnes. This can find open areas up to 2 meters/yards away.
I saw a sensor like that in Michael Gasperi site
(http://www.plazaearth.com/usr/gasperi/lego.htm). Maybe it's the same one.
I am trying to use Lego pieces only. Don't understand me wrong. I am not a
Lego purist. I simply don't have the time neither the eletronics experience.
I'm graduated in electronic engineering but I don't touch a resistor for
about 10 years. I always worked with software development . But if I find
some assembled device that I can connect to the RCX without the need of
oscilloscope and mulitmeters, I will. :)
I tried the sonar approach using the RCX IR emitter and light sensor. It is
not so effective, of course. I'm thinking about something using those lasers
pointers but I need some lens assembly. Some people discussed about this
aproach already but I saw no real application until now.
> Storing only object
> locations is best, IMO, because it uses the least space (there were posts • on
> this before..). Then you can just see if your current path intersects any • of the
> known obstacles.
I saw some posts discussing bitmaps vs. vector information. The object
location is a good option. My idea of storing paths is not really different.
It will only consider objects bigger than they really are. I have a special
problem with chair legs. :)
>
> P.S. I would love to help you. I have been trying to do something like • this for
> as long as Mindstorms has been out. You brought some new ideas to my • project.
>
Good. Let's keep the ideas flowing. As I said I will work on a slow pace and
post success, failures, code and building instructions.
Mauro
> Mauro Vianna wrote:
>
> > Jason,
> >
> > Of course a second brain will help!
> >
> > Everything is in my mind yet and lots of issues are not solved. I hope • this
> > post generates interesting discussions on how to solve them. As I said I
> > will test these approachs in a slow pace, most because of my lack of • time
> > than my will. I want to post all interesting features I achieve (maybe • even
> > a simple site later?).
> >
> > Although the project may seen a complex one, I want to test a lot of • smaller
> > tasks at a time, so that I can join them later as a bigger problem. For
> > example, I have tested the following features until now:
> >
> > 1) A sonar-like feature using the RCX IR emiiter and a light sensor to
> > detect IR reflections. It was described in this list earlier (not by • me). It
> > worked fine but wasted batteries fast. Some people gave me a few tips to
> > save them, but I did not test yet.
> >
> > 2) An algoritm to use light detection as a compass. The basic idea is • that
> > my window is the brighter light in the room (it only works during the • day)
> > so if you detect the brighter light source you can find a direction • without
> > a compass. Not my idea really. See the Ants project at MIT site. The • author
> > copied it from real ants too! Sorry, I lost the URL. :(
> > It's not precise and depends a lot of the environment but it worked well • in
> > my apartment since all windows are on the same direction.
> >
> > Unfortunatelly I don't know LegOS and in fact I don't want to learn it • right
> > now. I know the language is powerful is good but don't have the time • now. I
> > mentioned both LegOS and QC since those don't have the same memory
> > limitations of the original firmware.
> >
> > NQC is bound to the firmware variable limit, so map representation is • not
> > possible unless it is hard coded. One workaround I thought about was to • use
> > a simple NQC program that explores the environment and send information • to
> > the datalog. Then it returns to a location where the IR tower can reach • it.
> > A more complex Visual Basic program could process the datalog and create • a
> > NQC code to represent the map from the datalog info and download the new
> > program with the hard-coded map. The idea itself is simple. The main • problem
> > would be the lack of precision of the explorer info. RCX doesn't have • GPS,
> > compass, etc. so obstacle location would not be real precise.
> > Of course, we can simplify the problem. In fact I don't need to know • where
> > the obstacle is, but where it isn't. Once a path is found (e.g. straight
> > movement for 2 meters without obstacle), we can store only this • information.
> > A simpler map algoritm could be used. David Leper used a similar one in • his
> > site: http://www.geocities.com/Colosseum/Dome/5088/Hand.htm (This is not • the
> > root!)
> >
> > Mauro
> >
> > "Jason" <jasonrowoldt@yahoo.com> wrote in message
> > news:FyMqs8.9Bx@lugnet.com...
> > > In lugnet.robotics, Mauro Vianna writes:
> > > > I had an idea to a project that I want to start in a very slow pace. • In
> > > > fact, it is based in lots of ideas already presented here in this • list so
> > I
> > > > am not sure if someone already tried it before. Any help is welcome.
> > > > The main goal is to build a Lego robot able to navigate inside my • home
> > > > without assistance. Although it is simple, it involves a lot of • issues,
> > some
> > > > of then already discussed here. The main approach follows:
> > >
> > >
> > > <snip a really interesting concept>
> > >
> > > Wow, I would love to see documentation of the process you go through • here.
> > > Alone you might get discouraged with so complex a task. I got the • Mindstorms
> > > 1.0 a few months ago but have yet to really fool with it.
> > >
> > > This may be what jumpstarts my activity.
> > >
> > > Let me catch up with you, especially with LegOS, and I'll be a second • brain for
> > > this project if you'd like.
> > >
> > > Jason
>
>
>
>
>
|
|
Message is in Reply To:
 | | Re: Autonomous Robot
|
| For mapping, you can use an ultrasonic sensor, like the one designed by John Barnes. This can find open areas up to 2 meters/yards away. Storing only object locations is best, IMO, because it uses the least space (there were posts on this before..). (...) (24 years ago, 2-Aug-00, to lugnet.robotics)
|
37 Messages in This Thread:
 
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
