Hi everyone,
The question comes up from time to time as to whether the Pixhawk is a good alternative to DJI, Mikrokopter, etc for professional use in heavy-lift cinematography. So I wanted to mention this high profile and *very* high risk job we recently completed for Fox Sports as part of their coverage for the FIFA 2015 Women’s World Cup here in Vancouver.
We flew quite a lot over two days, racking up 4 hours of flight time in 19 flights, mostly over water, and much of it not only within controlled airspace but literally adjacent to one of the world’s busiest floatplane ports, right under the nose of the control tower. They had 396 landings/departures and 242 transits through the zone over the two days we were flying there!
You can see our flight paths at the main location in the front here, and two additional locations in the background:
The Pixhawk performed flawlessly, the only things getting in the way were pre-arm issues which although frustrating at the time are reassuring in that you know it is doing a good job of checking everything is 100% good to go before taking off.
One constant issue I have with the larger octocopters is that it needs to be perfectly still during startup or you get sensor calibration failures. This is easy to do on a small rigid quad, but on a large one sitting atop a 3-axis tripod gimbal with multiple layers of dampening, it tends to sway around a bit after connecting the power. I’ve mostly solved this by having some hard foam wedges to jam in between the gimbal and frame during startup, but it would be nice if this issue went away. I also frequently turn down jobs that would require launching from a boat because I know it couldn’t arm in that scenario. My suggestion is to have the sensor calibration converge over a period of time, so it would happen quickly in the normal case of a small rigid quad on solid ground, but take a lot longer to average out to a solution when there’s movement.
The other minor issue was compass failures during startup when launching near to metal. This is a minor one, and once you know about it it’s not a big deal but it caught me completely unaware and the failure message isn’t reported by the Tower app (reported) so I needed to get the laptop out and diagnose what was going on (while the Producers stood there tapping their feet!). In my case it was a concrete surface but there must have been a metal beam right underneath. It failed a half-dozen times before seeing the error on the laptop, but after moving to a slightly different location it armed right away.
Here’s some more info if you’re interested, I wrote a little puff-piece for a less technical audience on LinkedIn:
Thanks to all the devs (especially Randy) who make this such a great platform, keep it up!
Comments
Andrew, I totally hear you about trouble booting up an Octo and having is wobbling around while trying to calibrate the gyros.
On my large Octo, it is powered by a BEC which includes a fail-safe magnetic power switch. So I connect the batteries, but the Pixhawk remains unpowered. I then have a small pin to pull, and then the Pixhawk boots with little movement. Seems to work very well.
Tridge: sounds like a more thorough solution on both counts, good stuff
John: camera is a GH4, which is almost the perfect aerials camera in my book. The gimbal is customized from a famoushobby BG003 (added dual bars and aluminium roll bar, both to prevent any possibility of twist when reconfiguring). Harry at FamousHobby is *very* helpful -- it is strange that they don't have a better website and their gimbals aren't better known, but other companies sell them rebranded, e.g. AerialPixels
What camera/gimbal are you flying?
We'd like to eventually get rid of the gyro calibration step at startup. Instead it would run a filter (possibly the small EKF?) to estimate gyro biases over time, and arming would only be allowed once that has converged. That should allow reliable arming on moving platforms like boats.
For the magnetic interference Paul has done some work where the copter is picked up and turned around before takeoff, allowing it to get some time clear of magnetic interference.
Thanks for the report! We are looking at solving these types of issues.
Cheers, Tridge
Concrete is full of steel rebar and will impact your RF/mag devices. Here's what a concrete surface looks like before the pour.