Log in

Sun, Nov. 5th, 2006, 07:17 am
Killer Robots

Remember the awesome automatic Sentry Guns from Aliens?

Samsung has just built what appears to be the first commercially available version. I see that the wikipedia article on sentry guns has already been updated to link to them.

This goes on my wishlist along with the Toshiba Mini Nuclear Reactor.

Wed, Apr. 5th, 2006, 11:33 pm
Honda's Google Earth car navigation system

Unfortunately, it's part of their ‘internavi Premium Club' package.

I hate the idea of paying a monthly charge for these types of in-car systems. This kind of thing can't be hard to build yourself: just get a cheap in-car computer system, a GPS card, and let the car computer mooch internet off your cell phone. Should be possible for well under $500.

I want to build an in-car system, but I first want to find a cheap b&w 12v lcd screen. There's some pretty cheap color screens like this one, but I don't need color, and b&w should be much cheaper. But it seems like nobody makes them. I don't want to invest a lot in a screen that might quickly melt in the Arizona heat.

Sat, Mar. 11th, 2006, 01:30 am
Attack of the Killer Kites

As a kid, I always thought flying kites was a lot of work for a rather boring experience, but that it'd be really really cool if you were allowed to put blades on your kite and cut each others ropes and generally cause massive aerial destruction. I didn't know that doing this was actually a real sport somewhere.

Thu, Mar. 9th, 2006, 11:27 pm
Google - the second coming of Multivac?

Seeing articles tonight about Google's new calendering project and acquisition of the online collaborative word processor Writely reminded me yet again just how powerful and integrated Google's system is getting.

Many of Isaac Asimov's short stories deal a computer called Multivac. Multivac is a massive mainframe-style computer (sometimes The planetary computer), into which all data in the world is loaded. The operators of Multivac then ask it questions that it uses its massive database to answer.

Like Google's system, Multivac is enjoys a high level of trust by its users (In one story, Multivac decides national elections based on answers to seemingly random questions asked of a single 'representative' voter), and, like with Asimov's robots, Multivac (almost*) always deserves the trust.

I really liked the Multivac stories growing up, but I had given up on the idea of a monolithic, completely logical, brute force analysis computer system ever becoming that useful and intelligent, especially as I got further into my computer science studies and career. I considered systems like Cyc and IBM's Deep Blue interesting special case solutions, but useless as far as A.I. went, because the data required would be simply too large.

If we were ever to get intelligent computer systems, it'd have to be the same way nature did it, which meant software brains (probably neural nets) that learn by experience, not by programmed sets of rules.

But what I missed was the simply massive amounts of data that Google is sucking in. Google is building the smartest system on the planet, and they're doing it the Hard Way - by physically storing/digitizing all the data in the world. Their system is still very limited, compared to Asimov's Multivac system, but every year the range of questions Google can answer increases. They may just be able to do it.

Heck, maybe that old quote "I think there is a world market for maybe five computers." actually makes sense, if you're talking about monolithic Multivac/Google type computer systems.

* The exception is "The Machine that Won the War", and even there, the issue isn't Multivac's trustworthiness, just its reliability.

** I use monolithic in this post to mean that there exists a single 'instance' of the system (we don't each have a copy of Google), not that the system runs on one machine or is tightly linked.

Sun, Feb. 19th, 2006, 07:07 pm
United Arab Emirates joins the space race

From the New York Times:
"The vehicles would be designed by a Russian company, and the first ships could be ready before 2008, said one of the entrepreneurs, Hamid Ansari, who with his wife, Anousheh, and brother, Amir, helped finance the Ansari X Prize competition — the one that resulted in the first privately funded human flight to the edge of space, in 2004.

Space Adventures announced yesterday that it would build the spaceport in the emirate of Ras al Khaimah with an initial $30 million investment from its government. Eric C. Anderson, president and chief executive of Space Adventures, said Singapore would soon announce spaceport plans of its own."

This seems to be in competition with the planned space port in New Mexico that will be used by Virgin Galactic's SpaceShipTwo, but the article is rather light on details for the Russian-designed vehicles.

Wed, Feb. 8th, 2006, 02:25 am
First try at using lasers for home security

For around a month I had a laser set up to bounce through the rooms in our apartment at about knee level, that I was using to detect any movement through the apartment. It was set up to bounce off a couple mirrors and hit a simple detector in the 'computer room' that was polled by my main computer, which kept a log of all laser breaks (and beeped at each detected break).

This surprisingly didn't seem to annoy Loving Bride at all.

The laser break was also tied to my computer's video camera, such that when a break was detected, the frames per second of the capture would bump up to a higher rate, on the assumption that something more interesting was happening. If no motion or laser breaks occurred within a certain period of time, the frame rate of the camera was allowed to drop back down to only a couple a second.

I was hoping I could get something setup that I could leave running while we were gone for a week's vacation, but I kept getting problems with the laser drifting off the detector. Sometimes it would stay on target for up to a week, but often it would drift off within a day or so.

Both the laser and the detector are pretty securely fastened down, but I suspect the reflector that I used for the fine targeting adjustments slips just enough due to day-to-day vibrations in the apartment to throw it off. Towards the end, I was noticing that I was even getting spurious detections when planes flew overhead (we live less than a mile in the approach path to an international airport), and I think that extra vibration was enough to shift the mirror out of the way.

For this first attempt, I had just wanted to see how far I could get with a really simple system ($5 laser pointer and a photocell/555 timer detector circuit). I'm sure I could get it to work better if I was willing to get more serious about securing everything (bolting stuff to the walls, instead of just clamping and taping things), but I think next time I think I'll try to see about setting up something that auto-corrects when the beam drifts.

Mon, Jan. 16th, 2006, 08:07 pm
Airships for the Canada

This story advocates using airships for transporation in the Canadian north.

Fri, Jan. 6th, 2006, 02:09 pm
Mile-long airships being considered


When forest fires broke out in a remote area of Massachusetts a couple of years ago, firefighters hampered by a lack of access roads to the blaze relied on helicopters to douse the flames. But the helicopters were ill-equipped for the task: each small craft could carry only 100 gallons of water. In fact, to extinguish the blaze, firefighters spent several days scooping water from a pond, flying to affected areas, dumping water on the flames, and then returning for more pond water.

Mike deGyurky, a program manager at the Jet Propulsion Laboratory (JPL) in Pasadena, Calif., figures there must be a better way to handle such emergencies. His novel approach would entail deploying a giant blimp, perhaps a mile in length. With a capacity of 50,000 tons or more, such an airship could pour more than 10 million gallons of water on a forest fire. It could also be used to transport food and water (perhaps even a small iceberg) to drought-stricken regions, dump tons of dirt and sand on a nuclear reactor to smother an incipient meltdown, or even deploy a vast tarp to contain oil spills from leaky seafaring tankers.

Sun, Dec. 4th, 2005, 07:29 pm
PIC Project Progress

I had the prototype of my PIC project performing all the functions I wanted, so I've mostly just been doing cleanup and reorganization on the project during the last month or so.

I've been playing with the Debian version of Eagle to try to lay out a board for this. It's kind of fun watching the autorouter try to optimize the traces, especially when I make the job impossible by restricting it to a single-sided board. I'm hoping to have the whole thing to fit in a single-sided 2" by 1.5" or smaller board without having to mount components on both sides or something. Olimex seems to have decent one-off prices (about $1 per sqare inch) on this, and I'm trying to see how low a price I can get, even for a first run.

Yesterday I finished building a new, more compact prototype circuit from the eagle schematic. It was nice to have it just come together, and then work as expected once I plugged in the PIC. I made sure to double-check everything else before powering it up with the PIC though, to prevent a repeat of last time.

The new prototype circuit is a whole lot more compact than the previous one. Most of the new circuit fits on a single breadboard, whereas the first draft was spread across 5. The circuit itself is really pretty simple though, just the PIC, some optoisolators, some resistor networks and LEDs, and a 25-pin header for parallel-port interfacing.

I still need to get a version of this working with the PIC 12F675 chips. This project really only requires 4 I/O pins, so those chips would be a lot more appropriate than using the bulky 16f84a.

skipped back 10