Planning collaboration over the next several months

[dweiseth]

If anyone is still interested in fleshing out the remaining components for the robot let me know. I plan to work two projects in the next couple of months:

1. Basic Programming roadmap for the purpose of teaching Math reasoning & employing the power of computer to solve math problems in the real world.( need about 4 weeks before I have anything to demo )
2. Creation of machine learning code using probabilistic models enjoining time & space ( would like to test this in the robot if possible)

Anyone is welcome to use my phone again since neither of my projects require this in the next 4 months.

Cheers --David

[Paul Danset]

FWIW, I'm still working on a robot with Android as a brain controlling an Arduino.  I'm just doing it on a "back to basics" two servo robot; something that I can afford and work on at home any time.   Once I get it working I'll post the code on github and also help adapting it for our main robot.  Ideally the same code should run on both and accommodate less or more sensors.  My robot has no sensors right now other than what's on the phone -- no OFS, no sonars, external GPS.

Re: teaching math.  Have you looked at Sage?  It's like a free open source version of Mathematica with a Python back end.  I believe it was started at UW:

http://www.sagemath.org/tour-graphics.html

Years ago a local K-12 teacher was looking for an idea for a collaborative activity for kids on the net.   What I proposed was to have the kids measure the curvature of the earth by the following simple experiment:

• figure out the longitude / latitude of your location.
• at a specific time of day, measure the angles (inclination / declination) of the sun in the sky.  one might use a protractor or sextet here
• submit your angles to a repository.  this can be done manually or one could add a web CGI programming activity to create a simple form to collect data (latitude / longitude / angles) from different students / schools / states / etc.
• by doing some simple math (e.g. taking the average or weighted average of the data), one could deduce the curvature of the earth.

By knowing the position of the sun from multiple geographic locations, one can calculate the curvature of the earth (make the simplifying assumption here that the earth is perfectly round).  In fact, the more distributed (geographically spread out) the measurement, the better.

Figuring out the longitude / latitude as well as synchronizing the time for measurement is now even easier by simply using a smartphone.

So the kids will get some idea of math (simple trigonometry, simple statistics), geography, time zones and maybe some web programming as well.  And the kids would be collaborating with others across the net.  The teacher seemed really excited with the idea but then I never heard back.

Re: machine learning.  Teaching machine learning to young kids might be much more challenging, since you need a fair amount of math background to understand the algorithms.  Hand waving things like optimization / gradient descent only go so far.  (I did my masters thesis in neural networks and signal processing.)

-- You received this message because you are subscribed to the Google Groups Robo-Magellan Project  group.

Check out our project page: http://www.jigsawrenaissance.org/donation/robo-magellan-project/

To post to this group, send email to robo-magel...@googlegroups.com.
To unsubscribe from this group, send email to robo-magellan-pr...@googlegroups.com.
For more options, visit this group at https://groups.google.com/d/forum/robo-magellan-project?hl=en

[Pat Tressel]

David --

If anyone is still interested in fleshing out the remaining components for the robot let me know.

Yes.  I'm intending to port the LISP polygon path planner.  That would allow using a standard polygon-based map, with polygons marking the inaccessible areas.

Yesterday, Tyler and I got a start on debugging his Arduino code.
 

2. Creation of machine learning code using probabilistic models enjoining time & space ( would like to test this in the robot if possible)

Of which two popular models are various forms of Kalman filter (somewhat restricted in what it can represent, but correspondingly immune to overfitting), and particle filters (can represent irregular distributions, with limited accuracy, but subject to overfitting).  We could start looking at some of Dieter Fox's work.

Anything you're especially interested in?

Another option is to test in simulation.
 

Anyone is welcome to use my phone again since neither of my projects require this in the next 4 months.

It's your Arduino that I'm using, right?  I'm being Very Careful with it.  ;-)

-- Pat

[dweiseth]

Sweet on the sage stuff, ok let's pick a meet up time again on the robot stuff if you want, we can overlap where it makes sense. Thank you.

What day a week works, or does that make sense?

[dweiseth]

Yes no worries on the arduino that is what it is for, not sure on the best software approach, would like to collaborate, but you can independently determine your pieces, just hoping to learn and collaborate on what makes sense.

Is there a day of the week people are meeting on this at Jigsaw?

Thank you

[Budi Mulyo]

We mostly meet on Wednesday during SCRoW, David, and we heads to Wikispeed on Thursday. Last time we added Robotics Thursday and Saturday because of the Robothon. Too bad you missed out on MiniBob presentation by Xandan and ChessBig.com by Jerry last Wednesday. Hopefully they brings them again this Wednesday. I want to check out Felix (the MiniBob big brother) also.

Best wishes,

--

Budi Mulyo

206.571.8430

 

Jigsaw Renaissance is a learning and making community, a collaborative community dedicated to collective education and creation.

For more information about JR (www.jigsawrenaissance.org), please visit our wiki page at wiki.jigren.org/Starting_Classes or this page.
Contact us, so we can chat about scheduling something cool together.
Thanks!