Thursday, May 24, 2012

JournoDrone 2: Learning from the past, Looking to the future

JournoDrone One had an important mission: to be a drone journalism platfom that was "powerful, durable, transportable, affordable, upgradeable and supported by a community of experts." It became a pile of foam instead.

But that's OK. Drone development, especially at this state of technology, is a matter of trial-and-error. That's why myself and fellow developer Acton Gorton are giving it another shot. We are taking all of the experience, knowledge, and goals from the JournoDrone One project and starting again with JournoDrone 2.


In my final post about JD-01, I listed three major lessons that I learned from crashing and working on the drone. Those three lessons are now directives for JournoDrone2: 1) make it strong, 2) make it serviceable, and 3) make it hand-launchable.

To make sure the project is as relatively low-cost as JD-01, like JD-01, the new JD-02 is essentially a repurposed, built-up radio controlled airplane. With new requirements in mind, I selected an FPV-style RC airframe. FPV, for the uninitiated, is "First Person Video," or essentially mounting a camera inside the airplane from the vantage point of an imaginary, tiny pilot. The results can be dramatic and breathtaking. Take, for example, this FPV video taken from an RC airplane dodging trees and flying down a waterfall in the French Alps:

The decreasing size and shrinking costs of video cameras has essentially made FPV, and the RC community that has sprung up around it, possible. More advanced FPV setups even include on-board transmitters that send the video to a ground station. Sometimes hobbyists use video monitors or video-equipped goggles which enable them to operate the RC aircraft solely on the video feed.

In essence, these are drones. What separates them from military drones are merely range, the quality and cost of the electronics, sensors, and of course, explosive ordinance.

Most any RC aircraft can be made into an FPV machine by simply strapping a camera onto the front. In fact, that's how FPV began. Since then, RC companies have caught onto this trend and started making specialized airframes that have increased internal capacity and features that make them amenable to mounting FPV gear (front-facing camera mounts, for example).

For this project I selected an FPV frame that not only had adequate internal space for radio equipment, a camera, autopilot, batteries and servos, but also convenient internal platforms on which to mount that equipment. In this frame, the platforms slide into the fuselage and are secured with screws. This does not seem to be an ideal method of securing the internal platforms, as the platforms will probably need to be taken out and reconfigured periodically. But if that becomes an issue in the future, the screws can be replaced with bolts and nuts.

As part of requirement #1, the airframe must be strong. Most FPV airframes are made either out of foam or fiberglass. My experience with foam had left me to believe it is probably not an adequate structural material for a "work" drone (unless the operator is especially skilled or lucky), and most fiberglass models seemed heavy and difficult to service in the event of a crash (a violation of requirement #2).

The fuselage for this aircraft, however, is made of a hard-yet-flexible plastic. Hobbyists have used this RC aircraft and crashed it without having to make major repairs to the fuselage. Thus, it seemed like a good starting point.

Front bay of the JD-02 fuselage. Hook-and-loop straps are used to secure electronics to the internal platforms.


JD-02's Autopilot, developed and manufactured by 3D Robotics. A nickel is shown for scale.

As I said before, this drone is essentially a hacked RC aircraft with some added special equipment. In this case, as with JD-02, the added equipment is a camera for taking photos (which will later be stitched and corrected into photomaps), and an autopilot to fly the drone in patterns.

The autopilot for JD-02 is this ArduPilot Mega 2.0 (APM2) from 3D Robotics. It's powered by an open-source, Arduino-compatible ATMEGA 2560 running at 16MHz and has a pretty comprehensive suite of sensors to guide it. It includes a GPS receiver, a combo gyro/accelerometer chip with a sensor-fusion processor, a magentometer, a barometric pressure sensor with a resolution of 10cm, a micro SD card slot for "black box" or other data storage, and a hookup for additional Arduino 12C devices.

Pursuant to the goal of the first drone, this open-source hardware is tended to by the community, who have developed a convenient GUI interface to program and tune the autopilot. Perhaps one of the best features of this board for us at is not only the ability to use it on fixed-wing drones like the JD-02, but we also can plug it into our rotorcraft (JournoCopter) and use it to pilot those.


We've essentially got to do the following things to make this drone deployable:

1) Strengthen the fuselage
2) Assemble in basic RC configuration for flight
3) Test-fly the airframe in its basic RC configuration
4) Program, mount, and configure the autopilot
5) Flight test airframe with autopilot
6) Assemble and mount the camera system
7) Test autopilot with camera system
8) Develop a photomap from drone

From there, it's a matter of collaborating with other journalists to find a mission. We have a couple of ideas for drone missions in the Champaign-Urbana area, some involving a longstanding storm water abatement problem, and we hope to contribute them to the Knight-funded community news website More updates to follow.