Robot helicopter’s impossible tricks

“Controls” is the branch of engineering that deals with the regulation of moving things. Thermostats and cruise control are the obvious examples, but control systems show up all over the place: temperature control, power control, attitude control… anything that, if left untended, is likely to become a puddle, a crater, or a ball of flame. Behind each one of those regulatory control systems is some code, probably written by a grad student, that spells out how the system should behave.

I was once a controls grad student, and one of the things we learned is that, once you’ve figured out how to control something, it’s not very hard, mathematically speaking, to control it amazingly well. Crank up the gain high enough, and you can make a tractor tap dance or a jet skywrite in cursive. In theory. The situation is similar to the old Archimedes quote “Give me a lever long enough and I can move the world.” It’s a nice line, but I checked the catalog, and they’re currently back-ordered on those super long space levers.

From a practical point of view, to do amazing things with control design, you need fast, accurate sensors with very clean signals and fast, accurate actuators with very clean response. This makes for a system that is heavy, hot, power-hungry, and very very expensive. Or rather, it always did until now. Everything is getting so much smallerer and betterer that old-timers like me can scarcely believe what’s possible. This is the best time ever to go to engineering school. Look at what you can build. Behold the insane quadrotor built by a group (of grad students!) at Penn.

Pardon my French, but that’s some crazy shit right there. Every working controls engineer in the world is wishing they were starting grad school right now.

The tactical value of these systems is obvious. Imagine snooping spy-bats that can perch on walls and quietly eavesdrop before flitting away into the night. Here’s a nice piece on perching planes at Stanford.

Prepare to see an unparalleled rush of innovation in micro-aircraft.

Home-made UAVs

A UAV is an unmanned aerial vehicle. In the old days, a home-made unmanned aerial vehicle would be called a model airplane, or perhaps an RC (radio-controlled) plane. But the fancy-pants term UAV is well earned these days because of the amazing things amateur enthusiasts can do with them. It’s remarkably close to what much more expensive military UAVs can do.

For example, I read a fascinating article in Make magazine about the GPS-driven autopilot you can put into your kit. Sadly, the article is not online, but the article’s author Chris Anderson (who happens to be editor in chief at Wired magazine), runs a whole web site called DIY Drones. They’ll help you get started, and when you’re ready they’ll sell you an miniature open source inertial navigation unit that costs less than $100. That’s something that couldn’t be had for less than a hundred times that cost only a few years ago. By the time you’re done, you’ll have a device that can go spy on your neighbors. I won’t dwell on the point, but you can easily imagine many more mischievous uses for a cheap easily built spy plane.

If you make them small and nimble enough, you can fly indoors. Here’s a simple blimp that can pilot itself around your building, but the engineers at MIT have made something much niftier: precision-controlled helicopters. This video is especially impressive.

Here’s a Popular Mechanics article from last year about MIT’s indoor flight lab: Crash-Proof UAVs Fly Blind at MIT’s RAVEN Aerospace Controls Lab. I wish I had this stuff back when I was in grad school!

Helicopter strobing

If you only visited Aspen during ski season, you might be forgiven for thinking the place is snowy all year round. Similarly, a third grader might well imagine his teacher lives at school, since that’s the only place he ever sees her. Any time a periodic observation is synchronized with the event it measures, things get screwy. This movie makes the problem clear.

No matter how surreal it looks, this helicopter isn’t doing anything strange. The rotor turns at a certain rate, and the camera happens to be snapping frames for the movie at exactly the same rate (or a multiple of the same rate, but that complicates the explanation). So the camera is catching the rotor blades at exactly the point as they whirl around. As a result they look like they’re motionless. But if you were on the ground watching, nothing would look amiss.

Here’s another instructive video of a more prosaic system: the wheel of a bike.

A notable application of this kind of synchronization was the Red Baron’s machine gun. In World War I, if you mounted the guns where they were most conveniently operated by the pilot (just over the nose of the plane), you had the unfortunate side effect of shooting off your propeller. The first solution to this problem, introduced in 1914 by Saulmier, was to use armored blades, so bullets did minimal damage when, inevitably, they struck the prop. But the cleverer solution by far was strobing: use an interrupter gear that only lets the machine gun operate when the prop is safely out of the way. Just as with the helicopter video, from the gun’s point of view, the propeller is motionless.

X-36 pictures

The other day when I was trying to track down an old friend from my previous job, I ended up on a Beach Volleyball photography site. Because that’s what I used to do in my old job: pro beach volleyball. Yep. I was quite the pro beach volleyballer back in the day. You can probably find me in these action shots.


Actually, I worked with a guy who later went on to be a professional photographer of professional beach volleyballers. My real previous job was at NASA where I worked on a plane called the X-36. In the airplane business, you can spend years working on an airplane that never gets built. The good news is that they actually built and flew the X-36. Sadly for me this happened six years after I left the job. You can see from this picture that it was just a little guy, a 1/3 scale unmanned technology demonstrator. Because it was so small, it was often roughed up by the meaner planes at Edwards Air Force Base. Here it is being menaced by a gang led by the SR-71. That SR-71 thinks he’s so great.

Lots of other good pictures here.

But those are all the official NASA pictures. When did the X-36 take up beach volleyball? Here is a shot taken by VolleyShots photographer John Geldermann at the Induction Ceremony for the X-36 into the USAF Air Museum in Dayton, Ohio (July 2003). And those are the people I used to work with long, long ago.

Print that plane

People are starting to get used to the fact that unmanned aircraft, or UAVs in military parlance (for unmanned air vehicle), are being used quite a lot these days, particularly in Iraq and Afghanistan. Generally it’s in a nonlethal spying mode, but the occasional UCAV makes an appearance, where C stands for Combat. What’s counterintuitive about these vehicles is that, despite their moniker, they actually require more people for a normal mission than a manned vehicle. Another interesting tidbit is that, while there is no human on board the aircraft, there is in fact a human pilot. He’s just sitting on the ground at Nellis Air Force Base outside Las Vegas, 15,000 miles from the actual plane. Which is just amazing when you think about it.

UAVs have shown great promise, the most important of which is that they can complete a mission and never ever require you to send in a rescue team to recover a downed pilot. But they suffer from some shortcomings. First of all, the generals who buy them were all combat pilots, and they don’t much like turning pilots into videogame players. Also, they currently require too much manpower to operate. But this is beginning to change, and given the capabilities of current hardware and software these days, I’m sure it will change quickly.

One indication of this change is the Polecat project recently unveiled by Lockheed Martin’s secretive Skunk Works. Polecat shows great promise by simultaneously attacking the two great problems of any new airplane: the cost of building it, and the cost of operating it. Operationally the plane will feature advanced software that more or less allows you to tell it where to go without having to pay a fancy-pants pilot to step away from the craps table. Eventually these robot planes will unionize and drive up the operational costs again, but until then, we’ll be able to fly them damn cheap (relatively speaking).

Nicer than this is the fact that this plane was designed and built from scratch in 18 months. If we are to believe this, then aviation is entering a new golden age. Typical manned aircraft these days take a good fraction of a decade to develop. I was trained as an aeronautical engineer, and this one fact more than any other made me get out of the business. Throw the man out of the plane, and everything can happen faster. Beyond not needing seats and cup-holders, Polecat was built quickly because it was literally printed out by special 3-d rapid prototyping machines. In other words, the engineer who designed the wing could, after signing off on it, simply click a button that says “Make this now.” This is where the future is headed. Initially only R&D vehicles will be built like this. Eventually, though, your own customized car will be printed at a massive car printing facility near your home. You’ll be able to pick it up the day after you order it. Assuming the robot driver lets you get in.