phoenixgarage.org

It's been long coming, but I finally got my "Statement of Accomplishment" from Professor Ng at Stanford today. Many of us were waiting with baited breath; I was already sure of my score, but I was curious to see how the statement looked, and how it was scored...

I'm so late on this due to a number of reasons, but I thought I would combine them both in one article; to put it bluntly, Week 9 (Anomaly Detection/Recommender Systems) looked to be easy at first, but I got bogged down in the homework, while Week 10 (Large Scale Machine Learning/Application Example: Photo OCR) is looking to be a "gimme" week, though it started late.

Well, this was surprising to me: Despite this week being another "dual unit" week (something I was dreading, given my last experience with such a week), I actually found it to be fun, interesting, and anything but difficult! What were the units?

• Unit 13 - Clustering (K-means)
• Unit 14 - Dimensionality Reduction (ie, Principal Component Analysis)

Do either of those sound "easy" to you..?

Week 7 seems to have been the "easiest" week yet (ok, maybe the first week was easier?). One would think learning about and applying Support Vector Machines would be more difficult, but it didn't seem that way to me...

Well - this was an interesting week of fun and mayhem; I thought I had it all covered until I started in on the review questions...

Well, I "survived" another week of Stanford's ML class - Week 5 was all about back-propagation and how to implement it.

I frequently visit and read an atheist and science-oriented blog called Pharyngula. Not too long back, Pharyngula moved from ScienceBlogs to its new host at Freethought Blogs. Earlier this week, there was a server upgrade, and things got wonky; comments went missing, comments were assigned to the wrong posting, etc. PZ is understandably upset and exasperated. On that blog posting, one of the commenters, ColonelZen, posed a question that I thought I would try to answer, one that I have been asked in the past. My answer may not be the greatest, and it might act as a chainsaw where a scalpel is needed - but sometimes, drastic surgery is the only real option to getting the answers needed...

Between plugging away at finding a new job and working on the class, all-in-all, the week turned out ok - well, with the exception of our dog dying...

Well, as you know, this week was a major change for me; I won't go into any specifics there, but it did make me have to drop the AI class, unfortunately...

The past 15 hours or so have felt like a loose bicycle chain. This morning, that chain fell off.

John McCarthy, a distinguished computer scientist who coined the term "artificial intelligence", and was also the inventor of Lisp, has passed away at the age of 84.

Well - I'm still involved in both courses, and I haven't killed anyone yet, even though at times I felt like it!

I'm currently signed up and taking both of these courses, which started this past Monday:

Intro to AI
Machine Learning

Both complement each other as far as material goes, and I'm having fun, but this is all still clearly an experiment by Stanford...

Oh, to be able to own one of these for frivolous purposes...

Version 2 of this _MAPTRIANGLE cube for QB64 adds a few extra features...

This was something I whipped up tonight to show off QB64's _MAPTRIANGLE function being used to affine texture map a rotating cube.

I was doing my usual thing and clearing out the buffer of my ZoneMinder system, when I noticed a few of clips of the storm we had last night.

Analog joysticks are fairly easy to hook up and use with the Arduino. In this tutorial, I will describe the basics of hooking up such a joystick to the Arduino, and give some hints and tips on using the position values read from the potentiometers to control a basic vehicle using Ackermann Steering geometry, as well as to control a vehicle using Differential Steering geometry (like on a tank or bulldozer). Pictures and code provided!

UPDATE! We cut them open! Click here to see...

My wife and I have always had mutant lemons from our lemon trees (now bushes) out in our front yard - check out these new ones, the first new ones in a long time...

I'm not sure, but I think this is an old Cox 049 control-line model airplane engine...

In order for the Arduino I am using to command the steering motor to move left and right, I need to control the motor with a circuit known as an "h-bridge". Basically, this is a circuit that switches current to the motor to make it rotate in one direction or the other, at will (it also allows for easy speed control via PWM - but we won't need that here).

I'd already ruined one good motor (which ultimately turned into three), through the misadventure of driving the motor to the end of travel in the steering mechanism, causing binding and ultimately destruction of the gears. The only way to prevent this moving forward was to install limit switches.

In order for the robot to know how far and in which direction it is turning, it needs to have some method of feedback from the steering mechnism to tell it at what angle the wheels are. This is the basic principle behind all servo-mechanisms. For my Unmanned Ground Vehicle (URV), I decided to use a potentiometer for the feedback element. Properly wired, it can tell the UGV what angle the wheels are at (with a certain error percentage, of course).

The motors finally arrived, so I could now continue with the steering mechanism...

As I noted in the last article, I had only spent $10.00 on a motor that the original manufacturer wanted me to spend $350.00 for a bunch of gears, of which I only needed one. Don't get me wrong; the fact that they were willing to sell to me at all impressed me, and if I had a bunch of the motors instead of the three I did own (not to mention if they were new, and not well used like mine), it might've been worth it. Needless to say, though, their offer wasn't enticing, so I decided to do what anyone else in my position would do - SWAP, SWAP, SWAP!

Once again, its been a while (but not as long as last time!); progress is still being made.

Since my last post, I've been working on the front-end steering mechanism; let's just say either this is harder than it looked when I started, or I'm just not that good of a mechanic. Likely, it's both!

On another list I am on, someone pointed out this handy little guide on how to solder, written in a one-page "comic book" format. I thought others might find it useful, though I have a few extra points to add.

After reading this thread today:

Electro Tech Online: Waterfall Printer

...and the responses, and watching the videos; it got me to thinking how it could be implemented in as cheap a manner as possible. After thinking about it for a while (a few hours), I think I have found an idea on how to implement the valves, very cheaply (but there's going to be a bit of fabrication involved, TANSTAAFL); if you implement this, please at least throw a bit of credit my way, OK? Thanks. I'd build this myself, but I don't have the time or inclination. I honestly don't know if this system would work, but it seems to me that the idea is what is most important; I am sure someone out there could get this to work with some experimentation.

"It's like the universe screams in your face, 'Do you know what I am, how grand I am, how old I am? Can you even comprehend what I am? What are you, compared to me?' And when you know enough science, you can just smile up at the universe and reply, 'Dude, I am you.'" - Science Saved My Soul

It's been quite a while since I last posted an update regarding this project, but progress is still being made, if slowly. The last time anything was posted on this project...wow, has it been almost two years?! Ugh...

For a recent posting I wrote to a fellow member on the Arduino forums, I spent an hour or more tracking down a few historical papers on the technology of Virtual/Augmented Reality. My interest in these papers was mainly that of an amateur historical archivist; I had an idea that they existed, but in the past I wasn't able to successfully locate them. Either I became better at searching, or Google expanded its index; I greatly suspect the latter, actually...

Great news! A reader by the handle of Enif, on the Eletro-Tech (new window) forums, apparently has solved not only the pinout, but also the I2C protocol/command set it uses to read the buttons and write to the display...