Subject:
|
Re: DISTRIBUTED/PARALLEL CLUSTER for legOS through n*RCX
|
Newsgroups:
|
lugnet.robotics.rcx.legos
|
Date:
|
Wed, 15 Jan 2003 18:16:56 GMT
|
Viewed:
|
3034 times
|
| |
|
|
Last year I built a robot which used multiple RCXs in the way you mention. I
was building a sumo wrestling battle bot, and at some point ran out of sensor
ports.
For my sumo bot I'd built up some small abstraction layers around the sensors
and motors which made it trivial for me to add a second RCX and start using
it's resources without change the bulk of my program.
The way my system worked was more master and slave oriented. The original RCX
ran all the battle bot logic and was the master. The slave ran a subset of the
original code which simply watched for specific sensor events and when they
occured it sent the sensor value to the master over IR. A routine on the
master copied the slaves sensor value into a new global variable, which is the
same way local sensors are treated. I think I also had
motor control working. I used LNP over IR, and added my own simple RPC
interface for sending the events. I never tried to share the memory between
the RCXs, but I can see how you could easily build an interface which would
treat the other RCX like a file that you could read and write from. I don't
think trying to emulate RAM over IR would be useful.
In the end my battle bot code was rather large, and likely overkill but it was
fun to design. If you like you can e-mail me off list and I can share the
code. I think the last time I had it running was on a patched legOS 0.2.6. It
may be more trouble then it's worth to compile it, but reading the code may
give you ideas.
mark
|
|
Message is in Reply To:
7 Messages in This Thread:
  
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
