Subject:
|
Re: Challenge Ideas
|
Newsgroups:
|
lugnet.org.us.smart
|
Date:
|
Thu, 23 May 2002 23:57:03 GMT
|
Viewed:
|
1411 times
|
| |
|
|
I got the following in email from Larry. He also sent this to
lugnet.org.us.smart@lugnet.com but it does not seem to appear on the lugnet
smart page. Larry, you may want to check your registration on lugnet.
| Please don't laugh if I'm way off base here, I'm new to using RCX and am not
| so sure of it's limitations. I do have a challenge that I don't think anyone
| has addressed using RCX. For some time now I've been theorizing on how to
| develop a Lego RCX based platform that could accurately traverse every
| square inch of a given area. The problem is that the platform must know if
| it has or has not been at a certain location. This way, if an obstacle must
| be avoid at said location the platform can return and finish its task. This
| platform could then be built upon to do some simple tasks like vacuum a rug,
| sweep a floor, scrub a floor, dust a table top, cut your grass, and so on.
| Unfortunately I don't believe the RCX can store the needed data to remember
| where it has or has not been. I thought about using the RCX camera but I
| didn't want the platform tethered to an additional computer. Maybe a
| rotation sensor or two could be used to calculate how far the platform has
| traveled, but that means it must always start from the same location. GPS
| would be great, but I don't believe it is accurate enough at this time plus
| it is to costly. Maybe, this could be good for the next SMART meeting
| challenge.
|
| Thanks for letting me ramble on,
|
| Larry
I think you hit on two of the biggest obstacles with trying this on the RCX.
The RCX does have 32K of ram but it is only accessible if you use a non-Lego
firmware, such as LegOS. With NQC you only get 32 global integers and 16
integers for each of the 10 tasks. The other problem is that is it very
difficult to accurately keep track of where the robot is. You can do dead
reckoning if you use rotation sensors on the wheels. But it is very
difficult to keep this accurate for long, and, like you said, you will need
to start it in exactly the same place every time.
Still, I can see this as a good challenge. Given an arena of known (or
unknown) dimensions with walls, and unmoveable obstacles with unknown
locations, pick up all pucks/bricks/etc. that are scattered about in the
least amount of time. A variation to this will be to give the robot three
attempts. Between each attempt the pucks/bricks/etc. will be spread out
again. The best of the three attempts is the one that counts. This way the
robot can get a chance to learn the layout in the first two runs and
theoretically be fastest of the last run.
Gus
|
|
1 Message in This Thread:
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
