« December 2004 | Main | September 2005 »
January 30, 2005
Planet aggregator up and running
We now have a Planet aggregator running that collects up posts from everyone involved in CSCS (who has a weblog). Many thanks to Jon for suggesting it and Matt for taking the lead on getting it installed and set up!
An aggregator, in this context, gathers information. The particular information it gathers is a short blurb from a weblog. This way, as CSCS participants (who have weblogs) make updates to their site, it is easy for us to see all the changes in one place. This will help, we think, in coordinating the new Atomic Death Robot initiative. (Well, all the stuff we're doing with the Brainstem and Budget Robotics platforms).
The aggregator is running at cscs.cs-ed.org/planet. You can view it there. If you would like your weblog included in the list of blogs it tracks, drop a note.
The aggregator will update (at most) once per day. However, if your weblog is in the list of those being tracked, you can actually force the CSCS aggregator to update whenever you update your weblog. Here's how:
- In your weblog editing tool (or, from the web-based interface, if you post through the WWW), find the bit about "pings"; this part of the interface indicates which services on the WWW should be alerted to the fact that you've updated your page.
- Add to the list http://rpc.cs-ed.org/cscs/ping.cgi.
That should be it. If all goes well, whenever you update your weblog (and ping that script), then the aggregator will update as well, and your most recent post will be included in the list immediately.
Posted by mjadud at 03:29 PM | Comments (0)
January 27, 2005
Compiling C programs on the PocketPC
Getting C things to run on the Zaurus isn't too difficult, and Damian set up a nice cross compiler for us.
Getting Linux to run on the PocketPCs don't seem to be so easy though :(, so we are probably stuck with WinCE. However Rob got C# to work on the PocketPC, which is very cool.
But C# is not C, and thats what I need to get the Transterpreter running on the PocketPC, so a bit of googeling turned up this: Pocket GCC. If it works, which it should, there should be no/little problems compiling a PocketPC Transterpreter.
So in a while when we have the FFI on the Transterpreter sorted, we should be able to program both PocketPC and Zaurus robots in a proper language! Sweet!
Posted by clj3 at 08:24 PM | Comments (0)
January 20, 2005
The Kit handout
Tonight we handed out a lot of the kit, with the intent to get the brainstem project going. Some have to build (thankfully I avoided that!), others get to install Linux on some PDAs and I think Steve and myself got some stuff to just generally mess around with.
I spoke about my take on what I got and what I plan to do with it on my weblog.
Posted by rwatkins at 11:20 PM | Comments (0)
Things to do
Some possible things we might do next:
- Fun: Grab one of the First LEGO League tables, and try taking on some of the challenges ourselves. Gives us more time to get used to occam, and perhaps (*cough*) better at building.
- Fun/Important/Harder: We have a number of PDAs and some platforms for running little robots with PDAs as controllers. We haven't built these or got them running. We could take a few weeks and beat on these. I think, if we worked as a group, it wouldn't take long. The side effect of this is that we'd end up with new platforms, and we could then go back to #1 (if we wanted), or continue on with them. If all were to go well, they'd be programmable in C, Java, Python, and a host of other languages.
- Different: We could spend some time exploring StarLogo in a systematic way.
- Different: We could spend some time exploring Greenfoot, which doesn't work so well on my Mac. This is like StarLogo, but in Java, and very beta at the moment.
- Different: Play with XML-RPC, a cross-language remote procedure call protocol. Probably in Java
I might brainstorm some more ideas before tomorrow; we'll see.
Posted by mjadud at 01:52 AM | Comments (0)
January 19, 2005
Papers of possible interest
I (and others?) need to post more regularly to the CSCS weblog. Things that are cool. We should post them.
I'll automate some account creation this weekend, and see what happens.
For the meantime, I thought I'd point to two papers that, if you need some reading material, you may be interested in.
The first is titled A Robust Layered Control System for a Mobile Robot
, written by Rodney A. Brooks. This paper lays out the basics of the subsumption architecture for systems control. This involves layering more complex behaviors on top of simpler ones, so you end up with "reflexes" that can override more complex, planned behaviors. For example, you may be looking at something intensely, but if a bit of leaf blows in your eye, you automatically blink. This is because the blink behavior subsumes your conscious desire to (perhaps) read a sign.
The second is for those of you who are interested in concurrency and languages, and want something to wrap your head around. Jingle Bells: Solving the Santa Claus Problem in Polyphonic C# presents a solution to the following problem:
Santa repeatedly sleeps until wakened by either all of his nine reindeer, back from their holidays, or by a group of three of his ten elves. If awakened by the reindeer, he harnesses each of them to his sleigh, delivers toys with them and finally unharnesses them (allowing them to go off on holiday). If awakened by a group of elves, he shows each of the group into his study, consults with them on toy R&D and finally shows them each out (allowing them to go back to work). Santa should give priority to the reindeer in the case that there is both a group of elves and a group of reindeer waiting.
This paper isn't necessarily simple, but it provides an interesting problem that, if you're bored, you might try and tackle in either occam or Java. The former should be much easier than the latter, but either will be challenging just the same.
If you're keen, the Polyphonic C# paper cites an earlier paper by Mordechai Ben-Ari titled How to Solve the Santa Claus Problem. This paper compares a solution written in Ada to a solution written in Java.
While some of you may not enjoy your occam classes, the real world is becoming more and more concurrent/parallel every day. In March the Doctor Dobbs Journal will run an article titled The Free Lunch is Over: A Fundamental Turn Toward Concurrency in Software. It provides an interesting look towards the future and the role concurrency and parallelism will play. Learning occam (or, CSP, if you will) is going to serve you well.
Sorry for preaching there. Anyway. Enjoy.
Posted by mjadud at 01:49 PM | Comments (0)