Friday, March 2, 2012

The Harrisburg tornado, the JournoDrone, and a well-deserved crash

Aerial photograph of Harrisburg, IL. Not taken by JournoDrone One.

Today, the southern Illinois town of Harrisburg is morning six of its own who were killed by a 170 mile per hour, 200 yard tornado. It was one of the 16 tornadoes that ripped through Nebraska, Kansas, Missouri, Illinois and Kentucky, claiming 13 lives.

Tuesday night, while news networks came from far and wide to cover the devastation on the ground, I worked to prepare JournoDrone One to film the disaster from the sky. For better or worse, the drone never made it to Harrisburg.

JournoDone One is a test mule for, which myself and fellow drone journalism developer Acton Gorton hope will pave the way for a low-cost, highly durable and transportable small Unmanned Aerial System (sUAS) for journalists.

JD-01 had never flown a mission before, let alone been tested. So the idea of rushing to Harrisburg and filming anything was a long shot, but something I felt I needed to try. This was a disaster just three and a half hours’ drive from home base, and if I had anything to say about it, I would be putting my equipment to good use. And I do venture to natural disasters on a whim.

Of course, it wouldn’t do any good to drive to the storm-ravaged town of Harrisburg just to nose-dive a drone into the rubble after takeoff. The town had enough grief and did not need some clumsy pilot crashing drones into things and making things worse. So instead, I had a test flight in Champaign, where I dove the drone into a freezing field. Here’s what happened.


The GoPro Hero2 is a pretty nice piece of kit. It records in 1080p at 30fps, and weighs 5.9 ounces (167g) together with its rugged, waterproof case. It also comes with an assortment of straps, clips, and adhesive mounts. I found one strap in particular that seemed to fit the JournoDrone Perfectly (even if belly landings might not be the smoothest).

With the plane set-up for its test run, I assembled a field kit with two lithium polymer (LiPo) batteries, the transmitter, a bag of spare parts, various tapes and zip-strips, LiPo charger, LiPo balancer, and my trusty car battery jump starter, which can charge two or three batteries without being depleted. Should everything go smoothly at the test site Wednesday morning, I’d simply pack up and drive on to Harrisburg.

That didn’t happen. The forecast called for a 50-degree day and light winds, and I had not expected it to be 32 degrees with 15 mile per hour gusts in the test field. After fumbling with the drone, fingers now very much cold, the control surfaces were trimmed and the motor was responsive. I held the drone behind the camera, held throttle at 30 percent, gave it a chuck at a slight upward angle, and witnessed JournoDrone One immediately dive into the ground.

The wings flew apart (not necessarily a bad thing, as ejecting wings expend energy from the impact), and the avionics bay spilled out wires and electronics. Thankfully the HERO2, which is actually the most expensive part of the system, was unscathed.

I took a closer look at the damage in the workshop, and found a fracture in the middle of one side of the avionics bay. I fused the fracture using CA glue, and plugged remaining gaps with hot glue, and then patched the interior of the bay with fibrous packing tape.

The gray primer had begun to chip away and flake off, as the EP foam beneath it flexed during the crash. But this was OK by me, as the primer was just window dressing, and it can be easily re-sprayed later.

Factoring in repair time and an additional test flight, it was fairly certain that the window to survey Harrisburg was closed. The while Wednesday forecast there called for temps in the 50s and little wind, another storm front was supposed to blow through.

There was, however, time for another test flight in Champaign. And that might prove useful, so long as whatever doomed the previous test was resolved. Doing some quick research, I narrowed down the possible reasons for the failure:

1) Insufficient airspeed. The wings simply did not have enough air running past them to generate adequate lift. This may have been a result of inadequate velocity from the hand launch, insufficient throttle, or launching the drone in the opposite direction of the wind.

2) Too much weight. The more weight on an aircraft, the more lift required for flight. While the GoPro HERO2 is lightweight on its own, the case and mounting hardware may have tipped the scales.

I removed the GoPro camera and prepared for a second test-run. Fortunately for my fingers, the temps had reached the low 50s by the early afternoon.


The second test didn’t have much more success. The drone made more distance after being launched, but it did not “level out” and instead plunged nose-first into the ground. This time, the damage was more substantial; part of the nose broke off, the rudder and elevator separated from the fuselage, and the propeller somehow popped off.

Substantial repairs were required this time. Both the remaining nose of the craft and the fuselage were submerged in hot water and then dried with a hot blow dryer, which allowed the foam cells to slightly adjust themselves and regain some of the original shape. The separated piece was then glued back into original position, and held in place with rubber bands for several hours. Then the entire nose section was reinforced with fibrous packing tape.

Additional packing tape was used on the bottom of the fuselage to reduce friction on landing and add resilience. The foam panels on the underside of the wing which held the wing spar in place also were taped, as were the aileron servos.

Weight should not have been an issue for this flight test. This meant that only airspeed could have been the issue, and this may have been related to improper hand-launching technique.

Autopilots can allow drones to take off from the ground, but that generally requires a strip of concrete and landing gear. Catapults can be developed for drones that do not have landing gear, but that makes the system more cumbersome. For the time being, I believe that hand-launching technique will be important for journalists using small, fixed-wing drones, because it will allow drone journalists to carry less equipment, launch drones nearly anywhere, and do so with minimal setup time.

Below is a video which helps explain what goes into a hand launch:

The video provides some good advice, but fails to show a very basic part of the hand-launching process: gliding the plane without throttle to find the correct trim levels. It’s basically flying the plane in an unpowered glide, observing how it behaves, and adjusting the control surfaces accordingly. This is actually quite important as adding any throttle during a hand-launch will only increase whatever instability the plane already has.

 What’s the takeaway from this entire experience? Three things:

I) You can’t rush drone journalism. You can reduce preparation time with planning and experience, but you can’t take a drone from zero to taking aerial photos without proper training or testing. While it would have been fortunate to fly the drone at Harrisburg (and that may still be a possibility), this was a new rig that needed to have some bugs worked out.

II) Always add lightness. Small differences in payload, and how that payload is arranged on the drone, can make for a huge difference in stability. For your maiden voyage on any new craft, it’s best to fly with as little payload as possible. That way you can become used to the natural state of the craft and correct small imperfections before they become big problems.

III) Crashing is the hardest part of learning to fly, but don't let it slow you down. You’re probably not going to get everything right on the first shot, but you’ll never get any better if you don’t keep trying. And in this case, it was much better to goof up and crash in a wide-open field than in a town that's trying to cope with a great tragedy.

As for JournoDrone One, it’s ready for another flight (or crash), but the forecast for the next week does not look too promising. Here’s hoping that the week after brings warm weather, clear skies, and the lightest of breezes. For the drone, and for Harrisburg.