Exercise caution when flying in the Arctic

I just had an unnerving experience with my X-Star Premium, and want to put out a word of caution.

Last week, I was flying it in extreme northern Alaska, and encountered two problems:

1) In cold temperatures (but within stated tolerance of 32F/0C), the behavior of the drone can be erratic - even if you have warmed up the battery according to Starlink guidance. The most notable issue was that the video/picture controls are intermittent from Starlink - they are greyed out and it is impossible to start/stop video. It is *sometimes* possible start video while the connection works, and it will continue to record until you turn the drone off, but I had instances where it wouldn't allow video at all. In these cases, it seemed to not recognize the SD card as being present (?), so perhaps there's a temperature effect on some of the hardware connections? I had other intermittent issues as well, but this was the most obvious one.

2) I had a crash that resulted from the compass going out of whack during flight, leading to the dreaded 'toilet bowl effect'. In the image below, you can see the orange line (actual position) vs. yellow (commanded position) diverge during an arcuate path ending in a crash on the ground. This started while the drone was hovering. No geomagnetic storms were ongoing (there's a USGS magnetometer practically next door to where I was flying). I calibrated the compass a couple of days before, but had traveled several hundred kilometers since then, and the magnetic field in the arctic can be wonky (declination here is 20+ degrees, greater than you get anywhere in the continental US, and magnetic inclination is also very high at 85 degrees - perhaps the bigger problem). It's even worse near Antarctica, particularly in the southern oceans, but there's probably not a lot of drone flying there

I would suggest that Autel:
A) Increase the lower end of their operating temperature, based on real-world tests of their hardware
B) Consider a wider range of possible operating conditions (e.g. high declinations/inclinations?) and warn more frequently regarding compass calibration.


This image is from tomSny's excellent X-Star Log Viewer

I have not yet tested it since the crash, but I hope it's still airworthy!

 

What appeared to happen is that the compass/inclinometer flipped out (maybe reset in mid-flight?) which caused it to redefine what 'down' was to something about 10-20 degrees off of 'down'. This means that when the actual position diverged from the GPS position, the commands issued by the software could no longer bring it back into position and instead took it further and further away (which is why the paths diverge).

The official Autel line is "geomagnetic interference".

Here's the video, and you can see everything go haywire at ca. 30s in:
https://goo.gl/photos/rRts2wP5Rhp35uDa9

Unfortunately, the rest of the crash wasn't recorded, which is a real shame! Once it keels over, all controls were lost, and the video stopped recording. Interestingly, the power stayed on, and telemetry on the drone (but not on the Starlink app) was recorded - in the picture above, you can even see the track on the walk back to the road after the crash. It's the exceptionally odd behavior of a partial shutdown / freak-out that causes me to want to warn others - I don't really know what actually caused this behavior.

 

Sure, and it's a good point. The Kp index is a measure of how 'disturbed' the global magnetic field is. Basically, it tells you whether we're in a geomagnetic storm, and how severe it is. Lower numbers mean calmer, higher numbers mean progressively higher degree of geomagnetic storms.

This was one of the first things I checked, and I could be a little more granular than Kp index. There's a USGS Magnetometer in Deadhorse (exactly where I was flying), so I pulled the data from it.



I was flying at ca. 8pm local time, which would have been ca. 05:00 UTC on the 27th - as you can see from the graph, there was a bit of disturbance around noon the previous day, but for the most part, there wasn't much geomagnetic activity (I'd guess a Kp of 2-3?). The next day, however, there was a whopper of a storm!

NOAA has some nice dashboards for this here:
http://www.swpc.noaa.gov/content/space-weather-enthusiasts-dashboard

What you need to know for your drone is, if the "G" "Latest Observed" conditions are yellow or red - you might consider not flying!!

 

Be aware that the 'official' operating temperature range goes down to only 32F (0C).

I'm still suspicious that the cold was causing some kind of hardware problem, but it certainly may not happen in all devices!

 

That's unlikely - usually that's due to an unfavorable configuration of satellites and/or something blocking your view of the satellites (building, mountains, etc). However, it was probably smart to quit flying, especially if you aren't very used to manual mode!

 

Yes - I dumped the raw log file from the X-Star, so I have *all* the data. I can provide you a raw log file, or I've also dumped all values to a spreadsheet, so reading it into a GIS would be no problem. The only challenge is that the log values are by timestamp/by sensor, so if you want to compare GPS with Inclinometer, it will be necessary to impute the gaps. If you need help with this, just let me know. I'll try to clip out the relevant minute or so before the crash and send you a link this evening, and you can give it a go.

Humorously, Autel support said they had "no way of reading the log files"

I am a geologist, and I can tell you definitively that there was nothing underneath me but several kilometers of sediment - there should have been nothing locally magnetically anomalous at all.

 

OK - Here's the spreadsheet - I cut out only the bits leading up to the crash and immediately after. I have filtered it for "GPS" messages (magnetometer measurements would be tagged "IMU", for example), but they're all there. It's read only - make a copy as necessary.

I also did a quick chart of GPSTime vs. Altitude, so you can see the descent.

https://docs.google.com/spreadsheets/d/1QOrYfZMVvyP2HEKfPy5hcYAvSbcVlFMMcRmBMzzZBVA/edit?usp=sharing

The order of the individual message values are defined here:
http://xslogs.weebly.com/docs.html

I also dumped a KMZ file of the whole flight here:
https://drive.google.com/open?id=0By7tLN9aWlgadjBFaG1hSnVNUmM

 

Don't worry - there are piles of things I don't know anything about - and I'm a pretty novice-level flyer!