Robots

July 15th, 2010

I’ve been building little robots for quite a few years now, but my main focus has always been BEAM Robotics. Things have slowly been changing over the last year though. I started out by ordering some PIC microcontrollers, and while I managed to do a FEW things, I found myself hating them after a while. The problem with the PIC is that the only easy way to program it is through windows, and everything else is just annoying. Personal opinion maybe? In my defense, the programmer I had only allowed for a serial connection.

Everything changed though, at a recent Maker Faire. I picked up one of the new Arduino boards and a Getting Started book. After I got home, I was up and running within just a few minutes. It’s pretty amazing, actually. The Arduino guys have really done a great job in making a microcontroller that you can really use.

Anyway, back to the robots. I was recently at the local CVS and noticed a tracked, amphibious, R/C car. I bought two. I don’t know about anyone else, but if I show up with a shiny new toy and tell my son that it’s mine and I’m going to take it apart, it’s a little disappointing for him. So I took it apart pretty quickly and this thing is pretty cool on the inside. Everything has a rubber seal so that water can’t get in, and I really like the way it’s constructed. That being said, I had to open up BOTH of them to switch the motor wires… the factory managed to wire them in reverse. I guess that’s what you get for $14.99!

So I removed the top, and then went to work on getting rid of the excess electrics and plastic. I decided that there wouldn’t be enough room to mount everything I’d like, so I started modifying it to allow for a better top. But what top do I use? I looked at some hardware stores for the material that would work, and I figured I’d just build something custom out of balsa, when (tada!) I found an altogether amazing alternative. I have an old tupperware bowl that was meant for organizing or somesuch. The lid didn’t attach like a normal sealed container, but rocked back and forth. So without the lid, sitting upside down, it makes a great blue translucent cover. Sweet!

Next design move: I suddenly realized that I had an old NSLU2 sitting around that I don’t use anymore. The possibility of having a robot that actually runs Linux was almost too much for me! Whoa! Unfortunately, it wasn’t as easy as that. If I’m going to have Linux on board, I also want WiFi, and Python, and… a webcam, and… anything else I can cram in there!

SO… here’s the plan: I’m going to have an Arduino running as the interface for the motors and sensors, connected to the NSLU2 through USB. The NSLU2 will have Python running, which will act as the brains of the system. I’ll have a webcam and some sensors – I’ll probably start with just infrared range finders at each corner. Python will talk with the Arduino and make decisions based on sensor data, and will send movement commands to it. I’m thinking I’ll either setup an H-Bridge for the motors, or I’ll use some old servo boards that I have laying around.

The second phase will be to build out some software to all for getting telemetry and controlling the robot when it’s not running autonomously.

So much fun, right? I’ll update as I make progress!

Crazy-Cool Barcodes

November 12th, 2009

These are the coolest barcodes ever. I wish I made a product that I could put these on.

my first instructable

April 13th, 2009

Those that know me know that I’m a bit obsessive about creating. I love to build things, and if I go through a week without flexing my creativity, I really feel like that week was wasted. I’ve always assumed I’m a bit odd in that respect, but I’m not really sure. Given the amount of interest I’ve seen in Make: Magazine and their wonderful Maker Faire, maybe everyone out there is just a closet creator.

Anyway, because of my love of creating, my projects tend to hover somewhere between concept and almost finished. I’ll start on one, and then move to another and another before it’s done. My workbench is as you would expect… covered with gadgets at various levels of completion. With one of my latest project though, I decided to push through until I was finally finished with it. Yesterday I decided to finally just call it good (and stop tweaking it) and post it on instructables.com. So without further explanation, here you go: my first instructable.

python regex

March 25th, 2009

I don’t know why, but I never knew about the grouping that was possible in Python’s re module. I’ve been using Python this long and suddenly realize it’s there… a true facepalm moment. Anyway, how’s this for cool? –

[code]
import re
string = "S03E12 - A random filename"
print re.match(r'.*S(?P\d+).*E(?P\d+)', string).groupdict()
[/code]

Homemade steadycam

January 21st, 2009

I really like what this guy did to build a camera stabilizer out of what he had lying around. Exactly my kind of handiwork. I might try to make one myself…

stupid OS X…

November 13th, 2008

I recently started using a shiny new Macbook Pro at work, and I’ve mostly been in love with it. So much so, that I really do prefer the thing over Gentoo now. It’s just that things are so much easier. And hey, the prettiness isn’t anything to cough at.

There remains two things that annoy me to no end, though:

1) The hideous, insidious, freakious (?) lack of mouse tracking. Yes, I’ve read all of the writeups on why it can’t be done, and I’ve tried a few of the programs that claim to make it work, but none of them are really good fixes. They’re just… so so. I really wish there was some way to make it work. :cries:

2) That you can only resize a window with a tiny little handle on the bottom left corner! I came to the Mac from the wonder that is Enlightenment. I could hold alt, middle click, and resize from any region of the window. Now I KNOW this isn’t a problem with the way Darwin works. There are no logic issues. There’s nothing stopping Apple from doing it, other than a lack of time or pressure.

sigh.

One can still dream though.

music management

March 6th, 2007

After making the switch to linux full-time, I needed an easy way to convert my old music collection to mp3s. Yes, yes… there are the DRM’d files from my purchases on iTunes, and I won’t go into converting them. Otherwise, I had some files in mp3, and quite a large amount in m4a. It’s fairly easy to do the conversion in linux, but it can get tedious if your library is huge. Here’s a little shell script I wrote to parse a given directory and convert mp4s easily. I hope it’ll be useful to someone!

You’ll need faad and lame installed, natch.


#!/bin/bash

find /media/music -name '*.m4a' -print0 | xargs -0 -I {} bash -c 'FILE=`(echo "{}" | sed -e "s/.m4a//g")`; faad "$FILE.m4a"; lame -h -b 192 "$FILE.wav" "$FILE.mp3"; rm -f "$FILE.wav"'

exit 0

the wonders of vi(m)

January 20th, 2007

Ever since I started using Linux, which at this point was about 10 years ago, I’ve used the Vi editor. It was included as the default editor for most distributions back then, and it was really all that I was accustomed to. The usually cryptic commands make it pretty difficult to pick up, and I think that difficulty is why many people dismiss it so much. If you give it a try though, you’d be surprised at what you can do with it.

An article recently at The Register gave some interesting background. Essentially, Vi was made to be usable over a 300 baud modem. At that speed, even small commands are burdensome to type, and screen painting is horrendous. So the small, one-letter commands in Vi had a specific purpose – speed and efficiency.

So that begs the question: In the age of high resolution and broadband, why would anyone in their right mind still use an editor from 1976? The answer, to be honest, isn’t an easy one. Like I said, I personally started using Vi when I didn’t know of any other choice. I learned three commands… ‘a’, ‘:q’, and ‘:wq’, and that was good enough for me. Those commands allowed me to edit any file I needed to edit on my Linux boxes, and I could do it quickly. As any Vi user will attest, once you learn the very basic commands, you find yourself using them everywhere, even when you’re not using Vi. I think it’s because they are SO easy to use, they just make sense.

At this point, I suppose I should start saying ‘Vim’, for fairness. Vim was created to give a user the power of Vi, but extend it to add many cool features. The abilities of Vim are a surprisingly long list. Most anything you can think of, Vim can manage it, and it still remains a text-based editor. A GUI exists for it, sure, but I honestly know no one who actually uses it. :)

These days, I’ve had the wonderful chance to use Vim in my every day work, and I must say, I love it. I keep 3-4 xterms open all day long, and Vim runs on all of them. Something I’ve noticed is that like Linux itself, using Vim every day allows you to learn some incredibly useful shortcuts that you may never have seen. That’s my reason for writing this article… to educate those Vi users who only know the most basic commands, and to also help cement some of the useful shortcuts in my brain.

First, my (most recent) .vimrc file:
[php]
:colors slate
:hi Normal guifg=white guibg=black
:hi Search guibg=#224488
:hi Visual guibg=#111144
:hi DiffAdd guibg=#003311
:hi DiffChange guibg=#222222
:hi DiffDelete guibg=#110000
:hi DiffText guibg=#AA1100
:set guifont=Bitstream\ Vera\ Sans\ Mono\ 7.5
:set guioptions-=m
:set guioptions-=l
:set guioptions-=L
:set guioptions-=r
:set guioptions-=R
:set guioptions-=T
:set softtabstop=2
:set tabstop=2
:set mouse=a
:set number
:set expandtab
:set shiftwidth=2
:map :noh
:map :tabnew
:imap :tabnew
:nmap :tabnew
:vmap >gv
:vmap :vmap zf :set foldcolumn=5
:nmap zo
:nmap zc
:nmap :set foldcolumn=0
:map :noh
:map :s/^/\/\//:noh
:map :s/^\/\///:noh
:map :vert diffsplit =expand(“%:h”)/.svn/text-base/=expand(“%:t”).svn-base
:map :q:diffoff
:map :syntax match Character “\|”
[/php]

So some very useful things any newbie should know…

The ‘/’ character, in command mode, starts a search. This search accepts regex, and can be global or one line. So the simplest use is to simply type ‘/something’. That command highlights every instance of the word ‘something’ in the file. While still in command mode, typing ‘n’ will go to the next search result. Quick and painless, right?

The search and replace syntax is pretty standard. The replace the abbreviation ‘NY’ with ‘New York’, simply type ‘:s/NY/New York/g’. That will replace it on the current line only. The expand it to the entire file, just add a ‘%’ in front of the s: ‘%s/NY/New York/g’. Pretty simple, eh?

Here’s something that’s cool… split screen editing of the same file. Hit ‘Ctrl-w s’ and you’ll now have two copies of the current file on the screen. Do it again, and you’ll have three. ‘Ctrl-w c’ closes the current split. If you use the ‘mouse=a’ setting, clicking will switch between the different sections. Otherwise ‘Ctrl-w uparrow/downarrow’ will do it. There’s a surprising lack of split-file editing capabilities in editors these days. Enjoy it!

Did you know about Visual mode? Hit ‘v’ in command mode and use the arrows to highlight lines. You can then use ‘d’ for delete (like the ‘cut’ command), ‘y’ to yank (like the ‘copy’ command), and ‘p’ to paste. If you have the ‘mouse=a’ setting enabled, click-drag will automatically enter Visual mode. So it’s a quick clickdrag-y-click-p to copy and paste text between two places in the file. While you’re at it, highlighting a chunk of text in Visual mode and hitting ‘<' and '>‘ will tab/untab the block. 2007-02-02 – You can also hit a number before you hit ‘<' or '>‘ to tab that number of times.

One thing I like to do is store commonly used methods in regular text files. If I need those methods, I can insert them directly into the current document with ‘:r filename’. That reads the entire file in and places it wherever your cursor was located in the file.

If you don’t like the default colors that Vim uses, you can modify them. Take a look in /usr/share/vim/vim70/colors. Those are the default color schemes that come with Vim, and you can use them with the command ‘:colors [colorfile]‘. You can also just use them as a starting point and include custom color commands in your .vimrc file.

I know there are more things that I do in Vim that I can’t think of right now, so I’ll come back and update this article now and then. I hope at least one or two of these functions were unknown when you started reading. If you’re a Vim master and you’ve read this fafr, what are your favorite features? Please leave a comment!

Added: 2007-01-22

“$” means ‘end’. It can mean the end of the line, or the end of the file. Typing “:$” goes directly to the last line. And on the subject of moving around in the file, use ‘:’ and the line number to warp directly to a specific line.

Added: 2007-02-03

So recently I’ve started using (gasp!) Gvim, the ‘GUI’ for Vim. The thing that sold me on it wasn’t so much the GUI features (I’ve disabled them all) but the font rendering features. For one, you can use true-type fonts, rather than being stuck with the one in xterm. The other thing though, is that it does AA on the fonts, which means dot values! So I’m now using Bitstream Vera Sans Mono 7.5. It doesn’t seem like much, but try that in an xterm! :)

While I’m here, I’ve started using yet another feature. While in insert mode, with mouse=a enabled, double-click on a word. It enters visual mode and highlights the word. Now hit ‘d’ for delete, and it deletes the word and returns to insert mode.

Also, instead of using the method I detailed above for split-screen editing, I’ve started using :split and :vsplit. This allows you to split the current screen up quite a lot. Then, while in one of the split frames, type ‘:e path/to/file’ and you now have multiple files open in the same window. That was HUGE to me! And since they’re all within the same session of Vim, the copy/paste buffer will go between all of the files. That rids me of the need for the mouse=c functionality. Woo!

Added: 2007-02-13

It’s funny how your methods change over time. The amazing thing is how often I find new ways of doing something in Vim. I explained somewhere above how to use Visual mode. Basically, if you have mouse=a set, you can just highlight characters and you’ll automatically be in Visual. Here’s something cool though: try moving your cursor to the middle of a line and hit Ctrl-v, then use the down arrow. You’ll notice that instead of highlighting the entire line, it allows you to highlight a certain character region on your screen. This is Visual Block mode. Imagine if you had two columns of text and you wanted to move the second column a couple of tab stops over… Just put your cursor at the beginning of the column you want to move, Ctrl-v, down arrow to the end of the column, and then hit ‘>’. Neat, eh?

Sometimes you need to replace characters in a certain chunk of code, but not the whole document. Using either Visual or Visual Block mode, highlight the area you want the ‘replace’ to affect. Then you modify the normal pattern “:%s/something/something else/g”, adding “\%V”. So to replace ‘Yay’ with ‘w00t’ in a Visual area, you’d use “:%s/\%VYay/w00t/g”. Easy as Pi. ;)

moving to another city, and gentoo

January 13th, 2007

I’ve moved! I’m now in the amazing city of Austin, TX and into an equally amazing job. I love it, and the people I work with! One of the things I needed to get used to at the new place was using Gentoo on the desktop. Up until then, I had mainly standardized on Ubuntu in various flavors. I love the ease of use, and I love how… well… pretty it is. But coming from both camps, I wanted to give a basic breakdown of the differences, and why you might want to give Gentoo a try yourself.

The biggest difference between Gentoo and most other distributions is that every package installed in your system is compiled from scratch. There are benefits on both sides of the argument. In a binary distribution, the speed of installing apps is just amazing. Using apt is just… wonderful. The problems come in when you want to install an application that isn’t in the apt repository, or when the version you want to install is much newer than the one in apt. At this point, your choices include either waiting a while (possibly a long time) for the version to get updated, or compiling your own. Simple enough, but what happens when you go to install an application that depends on the one you compiled? Apt doesn’t see the compiled version in it’s internal database, so it tries to install it’s own. So your choice there is to install the other application from source as well. Don’t get me wrong, apt is awesome. It’s just that in certain instances, it would be nice to have a little more control.

In addition, think about compiling. Every application you compile has certain flags that enable different functionality. It might be a choice between using ALSA or OSS, or whether to let the app have SVG support… there are a million things. Well, in order to be all things to all people, the application in the apt repositories are compiled to work with pretty much everything. The side-effect of that is a potential for slowdowns. When you think about it, it only makes sense that the more you allow your program to do, the more CPU cycles it will take in running. And while one itsy bitsy flag enabling some random feature might not make a huge diff, what about scaling that across an entire operating system? If you have ever compiled your own kernel, it’s the same reason you don’t compile in every feature possible… you pick and choose for the most efficiency.

So those are the main reasons why I’m quickly becoming a Gentoo fan. I love Ubuntu for people who are new to Linux, but for those of us that know (or at least should know) a little more about our OS, Gentoo is a great fit.

Gentoo’s package management system is called Portage. It consists of what are called ‘ebuilds’, which are basically just scripts that tell the Portage system how to compile a given application. So for every application that is available, there’s an ebuild for it in the Portage tree. But there’s more! Remember the version issues you run into with binary package managers? There are ebuilds available for many of the latest builds of various software, and nearly every version number in between. So by changing a few configs, you can install an older more stable version, or you can allow Portage to see and install some pretty bleeding-edge stuff. Not only that, but there are some builds that actually use CVS to bring down the absolute latest version, and compile it automatically.

It’s hard to complain about the selection of software available on Gentoo through Portage. It’s got pretty much everything and the kitchen sink available for installation. And the great thing is, like all package managers, it understands interaction between different software. That means if you go to install WhizBang package, and WhizBang requires the SupaSecret KungFoo package, Portage will compile SupaSecret first, and then compile WhizBang. Not only that, but there’s even a program that will rebuild all of your dependencies, throughout the system, in case one gets hosed.

Binary package managers have methods (sometimes) for upgrading the system to the newest release. With Gentoo, the entire system is in a sort of continuous development. That means that there’s no specific ‘version’ of Gentoo… you install it, and then compile the packages (whatever the newest versions are) that you want. It’s very customized. That means that it’s very easy to bring your system up to date… “emerge –sync && emerge -update -deep world” will go through and find everything in your system that’s out of date, and update those packages. Swanky.

In all fairness, the advantages DO come with a price. The main thing that people complain about with Gentoo is the time that it takes to compile. For example, I installed it for the first time by myself on my laptop. It’s an AMD64 3200+ system, and it took slightly over 15 hours of compiling before the system was really at a usable state. By usable, I include compiling X, Gnome, the KDE libraries, WiFi drivers, and the various software I normally use. At that point, I was able to start really going through and finding other applications that I wanted… I’d say, all things considered, I’ve probably spent around a week of the last 1.5 months compiling. Of course, a lot of that was just me getting used to the new system, playing, compiling, removing, etc. From a blank drive, the process would go a lot quicker now.

To be honest, when I first thought about having to wait for EVERYTHING to compile, I didn’t like it. I was accustomed to the speed of binary managers that just download a compressed file and run an installer. Now though, I think I’ve adopted a slightly different view. Once the base system is done, compiling packages here and there really doesn’t take much time since the dependencies are all already fulfilled. In short, I think the advantages are well worth the wait. If I’m about to install a program, I just have a synapse that fires telling me to go grab a cup of coffee while I wait. Not hard at all… a little more casual, in fact. While I certainly don’t look forward to the time investment of installing another Gentoo system, the stability and efficiency is such that I don’t think I would be happy any other way. And you know… there’s also a big difference in the size of the OS… compiling everything individually keeps you from installing useless software that you’ll never even look at. As of this writing, having nothing else on my list (at all!) in need of installation, and with a few things that could be removed, about 8 gigs are used. That includes data files and all. Compare that to a base install of a binary distro, or even (gasp!) Windows.

So there you go. If I’ve inspired you to try Gentoo, please let me know. If I can help you over some of the beginning hurdles, leave a comment and I’ll get in touch. If not, maybe after a few years of getting more familiar with your Linux distribution, you’ll decide it’s worth it. If nothing else, I hope this has been educational.

Happy Desktop-Linuxing!

pseudo-realtime reporting of agent status in asterisk

October 23rd, 2006

Once you have a few agents on your asterisk box, and especially if they are distributed throughout the building, you quickly find yourself in need of a way of easily seeing who is or isn’t logged in. This helps with agent tracking, but it’s also useful for those times that you’re debugging your dialing stategies. So we’re going to write a Python script that will run constantly and update a database with all of the agents’ names and their current status. I’ve personally been running this for months with no issues as far as CPU load or instability in the Asterisk system.

First of all, you need to enable the manager interface. Look in /etc/asterisk/manager.conf. You’ll need something to the effect of…

[general]
enabled = yes
port = 5038
bindaddr = 0.0.0.0

[joe]
secret = schmoe
read = system,call,log,verbose,command,agent,user
write = system,call,log,verbose,command,agent,user

So Asterisk will be listening on the localhost address, on port 5038. Our user ‘joe’ is set up with the password ‘schmoe’. Now we need to connect to the server using Python. Here’s the script:
[python]
import socket, time, MySQLdb

def main():
db = MySQLdb.connect(host=”dbaddress”, user=”dbuser”, passwd=”dbpass”, db=”dbname”)
sql = db.cursor()
agents = {}
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.connect((“localhost”, 5038))
while 1:
try:
s.send(“Action: Login\r\n”+”UserName: joe\r\n”+”Secret: schmoe\r\n\r\n”+”Action: Agents\r\n\r\n”)
time.sleep(1)
data = s.recv(30000)
for line in data.split(“\r\n”):
if line[:5] == ‘Event':
if line[7:] == ‘AgentsComplete':
break
if line[:5] == ‘Agent':
agent = line[7:]
if line[:4] == ‘Name':
name = line[6:]
if line[:6] == ‘Status':
if vars().has_key(‘agent’) and vars().has_key(‘name’):
if not agents.has_key(agent):
agents[agent] = “”
if line[8:] <> agents[agent]:
agents[agent] = line[8:]
print name.ljust(15), agent.ljust(10), line[8:].ljust(20), time.ctime()
sql.execute(‘replace into ACD (agent, status, name) values (“‘+agent+'”,”‘+line[8:]+'”,”‘+name+'”)’)
s.send(“Action: Login\r\n”+”UserName: manager\r\n”+”Secret: manager\r\n\r\n”+”Action: Queues\r\n\r\n”)
time.sleep(1)
data = s.recv(30000)
for line in data.split(“\r\n”):
if line[:16] == “4 has”:
calls = int(line[line.find("has ")+4:line.find(" calls")])
if calls > 0:
print calls
sql.execute(‘replace into ACD (agent, status, name) values (“CALLS”,”‘+str(calls)+'”,”CALLS”)’)
except:
s.connect((“localhost”, 5038))

s.close()
[/python]
As you can see, I’m using sockets to connect to the manager interface. You can actually open up the interactive interface to Python and type these commands individually if you’d like to see the output that you’ll receive directly on the socket. Suffice it to say, the FOR statement starting on line 14 checks each line, looking for specific attributes that give us various info. In my case, I wanted to get the Agent ID, their Name, and their status. Once I have that information, I just insert it into a table that I created called ACD.

Something else that I should note is the FOR statement on line 33. This chunk writes a line to the same MySQL table with the total number of unanswered calls in the queue. This number allows me to know if customers are waiting unnecessarily long to get to an agent. You could obviously extend something like this to include timestamps, etc… allowing you to know average wait times and whatnot.

That’s basically it! At this point, you should install the script on your asterisk box as a daemon. Otherwise, you’ll need to keep a live SSH connection open specifically for this. Keep in mind, any time the Asterisk server dies for any reason, so does the manager interface, and your script. You’ll need to either restart it manually, or let your daemon restart it.

If you can think of ways to improve this script… I’m sure there are many ways… please comment below!