802.15.4 radio, i.e. ZigBee / XBee is a great new technology and I want to play with it! I also think it would be fun to be able to control my space heater, electric blinds, and desk lamp from the internet. So, I'm plan to buy myself some XBee radios, a USB controller for one, some relays and try and put together such a system. I'll start with the space heater so that before leaving my office at New Mexico Tech I can fire up the heater via the internet and come home to a nice warm room. Next I plan to set up the blinds and maybe put those on a cron job to open and close them to make sure I get sun in during the day. A more ambitious project is motorizing my desk lamp so it swings over towards my bed for bedtime reading and back to my desk without me having to get up and do it manually.

For several years I've dreamt of building a blimp to assist with cave exploration. Mapping caves in Austria with the Cambridge University Caving Club, we would often encounter "avens." These vertical shafts opening into holes in the ceiling usually require lots of arduous, technical caving to ascend. Since most avens are at least partially out of sight (commonly the ceiling is high enough to be out of view) there is no way of telling whether an aven leads to more passage until the hard work of ascending it has been accomplished.

Using a ballon containing buoyant gas attached to remote-controlled fans, a telemetered video camera, and a light, it would be possible to quickly scout these avens and decide whether they are worth ascending.

There are many challenges to overcome when constructing such a blimp. Since the radio connection will require line of sight, the blimp will be unable to go around corners and explore side passages unless some form of repeater is used (perhaps a second, "repeater blimp"). Transporting the blimp to the target aven is likely to involve narrow / crawly passages, in which case the blimp should probably be inflated at the target site from a compact bottle of compressed helium.

This is likely to be my PhD project, after I finish my master's! A pilot project has been started; see the physical characterization project for more info on this.

I plan to create a webpage with a canonical URL of www.aarongcurtis.com . It will serve as the center of my presence on the internet. There will be:

• A journal, with spatially and temporally referenced events.
• An opinions section, where I can spout off about stuff and people can argue with me.
• A whereabouts section, with my google calendar and an openlayers map of all spatially referenced stuff.
• A whoabouts section which lists my other online presences and contact info.
• A photos section.

I wanted to slice up [[image: howdoyoufeel]] for use on this website. GIMP has two useful features, guides and the guillotine, which almost do this for me automatically, but not quite. When you set guides on an image, guillotine will chop it up and save it as a different file for each rectangle. However, I want to take a 10 by 7 grid in a single image and turn it into 70 images. I could set 7 horizontal guides and 10 vertical guides by hand, but this would take a while.

So, as I usually do, I'm spending an even longer time writing a python script to do it for me. Probably a net loss of time, there is a corresponding gain in skillz and besides if I ever want to chop up an image again it I can do it instantly.

For those of you accustomed to modern LiDAR systems, this might sound impossible.

For those of you who aren't, LiDAR stands for Light Detection and Ranging and is a system where objects or landscapes are scanned with lasers to produce digital 3D models. A typical ground-based lidar system today weighs 20 to 40 lbs, costs half a million dollars, and contains fragile moving parts which must be carefully calibrated to produce useful data. Having said that, they are capable of collecting thousands of points every second with incredible accuracy.

I'd like to have a LiDAR system that I can carry in my pocket, for imaging cave chambers and passages. I don't need kilometer range or millimeter resolution, and I certainly don't have half a million dollars.

It turns out there's a device out there that almost fits the bill already. This is Beat Heeb's DistoX, http://paperless.bheeb.ch/ . Using an electronic compass, clinometer, and laser rangefinder, it knows where it's pointing and the distance to what it's looking at. It can store 2,000 readings or transmit readings to a PDA via bluetooth.

Beat's been kind enough to send me the firmware source code for the PIC on the DistoX, and has suggested how I can modify it to make the DistoX output a continuous stream of data. Once that's done, lidaring a cave room will simply be a matter of attaching the DistoX to a (non-ferromagentic) pole of some sort and manually sweeping it around the room.

I just set up a HOBO datalogger with a CO2 logger hooked up to one of the carboys to monitor our homebrew operation. It occurred to me how nice it would be to have wireless telemetry (poetry!) from the logger.

HOBO dataloggers, particularly the U12-008 tend to have extra space in their enclosures. I bet it wouldn't be too hard to stick an Xbee module in there and solder it up to some test points to wireless-enable your HOBO datalogger.

Hopefully I'll get around to doing this sometime.

Web site administrators often record visitor's mouse positions and clicks, usually as a diagnostic tool to debug the user interface. The admin can literally form a picture (often using a colored image called a 'heatmap') of what on the page draws user attention, when they become confused, etc.

I thought it would be a fun project to seek out alert administrators by gently spamming their heatmap systems. A crude way to do this would be to write a script that would go to randomly selected websites, take control of the mouse, and "write" an email address or tinyurl on the page in mouse movements.

A better way would be to find a lower-level way of triggering mouse move events to produce the same effect without actually using a fully functioning web browser.

Here's an idea for Google labs, which they've probably already thought of. Just in case they haven't:

You could publish an automatically generated list of topics for which the web fails to satisfy user demand. A simple way to do this would be to check which frequently searched terms have disproportionately low numbers of results. Another would be to check which search terms the following even occurs most frequently: a user makes many similar searches and / or keeps trying different links for the same search terms.

Publishing such a list could inspire web entrepreneurship, direct investments in wise directions, etc.

It would be great to have a firefox plugin that mutes Pandora when the ads come on. The problem is detecting ads, but I suspect something can be found in the DOM to indicate the presence of a a new ad. An interesting challenge.

Clark and I got a Kenmore mini-fridge from an Albuquerque Craigslister for cheap. It ran ok, but we soon found that the thermostat was overambitious -- even at the warmest setting it was keeping the fridge near freezing. We pulled the control unit apart and decided to replace the thermostat.

I thought this would make a good first PIC project for me, and so I set about building a thermostat using an LM335 temperature sensor, the PicKit2 low pin count demo board, a PIC16F688 microprocessor, and a big relay.

Having a digital programmable thermostat opens up many possibilities, particularly for brewing. For example, when lagering it's necessary to change the fermentation temperature at set points during fermentation. With a PIC in charge, it should be easy to program the fridge to change temperatures on specified dates.

I want to try increasing the temperature gradient to see if I can drive crazy vigorous convection.

Temperature profiles were collected in Warren Cave, a fumarolic ice cave on Mt Erebus. I need to analyses this data for spatial and temporal trends and patterns.

NOT YET PUBLISHED

Following the success of the <a href="/project/pic-mini-fridge-thermostat">PIC-based mini-fridge thermostat</a>, we here at the Allen Court Homebrewery decided to expand the operation, acquiring a full-size Maytag MTM1943ARW. After shelling out $50 to the emigrating neighbors we wrastled the fridge out their door, down the street and through ours.

The next step was to turn the fridge into an incubator capable of maintaining two 5-gallon carboys at a given fermenting temperature. These can range all the way from 7C (45F) to 32C (90F), but we mostly make ales, which are fermented around 17C (63F). After frustration stemming from incompatible code between the many PIC compilers I decided to use an Arduino instead. Much hacking ensued,

I would like to have the following workflow for most things I do (work and personal projects):

1. Plan an action using something like gqueues (sometimes with a date / time attached, sometimes not) 2. Record what I am working on, when (KDE activity logger?) 3. Publish everything useful on my website (hg browser)

Chances are that you, and most of the people you know, spend tens of hours in an airplane every year, sitting next to strangers. In those hours, you may have been on a plane with a long lost buddy, and not noticed. Flying with strangers, you might have had an interesting conversation or even made a meaningful connection with someone, but mostly you are lucky to make a "single serving friend." Quite often, you're probably next to someone you find mildly unpleasant, or worse.

I propose a way to fix this problem: a system of booking flight seats integrated with social networks, allowing passengers to decide where they sit, and also what flight they're on, based on the other people on that plane or in the seats nearby. Possible uses: - Group bookings (business travel, family) - Air dating - Avoiding a potentially unpleasant cabinmate

To begin with, I intend to implement this idea as a Facebook application. As passengers book flights, they may elect to share their flight information and seat location with a subset of their friend list, or possibly all of their Facebook friends.

To be truly successful, this idea would probably require airline cooperation and integration with online booking systems. Existing airlines are unlikely to get involved in such a risky paradigm change, but a new airline could capitalize on the concept. It is easy to imagine a succesful marketing campaign.

A dreamcatcher woven from core memory.

A plasmoid that displays these, updated automatically: http://sohowww.nascom.nasa.gov/data/realtime/realtime-update.html

When downloading dataflash logfiles into the APM Mission Planner, it would be nice to be prompted for notes about each flight.

Relying on the compass for direction means there will probably be some drift. My simple mode appears to always be a few degrees off after a few minutes of flying. It would be nice to be able to flick ch 7 to re-set the direction which counts as "forward" in simple mode.