PSA / WARNING: This Can Happen To You

I used to scoff at fly-aways too. I'm not so sure now...

Today was a holiday Monday in Canada. I was as at the construction site for a new dome field house soon to go up at the edge of a rural community. I'm part of a volunteer group documenting its rising. However for today, other than my son and I, the construction site was empty. It was very windy with strong gusts.

To make a long story short, my P2V+ was in a hover about 400ft away at 150 feet and at the far corner of the construction site. It was pointing almost at us. I was just tweaking the pan to take a good still photo when the P2V+ just took off. I mean it suddenly accelerated at full speed horizontally (the nose was dropped) in what seems to be a course perpendicular to the line drawn from my position to the location where the P2V+ was hovering. Full control deflection didn't effect it in the least. I quickly cycled S1 and S2 and left S2 in OFF (up) and S1 in GPS (up). I regained control for a few seconds and then it started twitching about the sky and then suddenly entered a near vertical dive. I cycled S2 and S1 again, put S1 into ATTI, S2 off and gave it full throttle. It came out the dive, I got it somewhat stable, and then I moved S1 to the Fail-safe position (full down) and it stabilized and, after several seconds, started to fly back to the home point.

When it was about 100ft away I moved S1 to the GPS mode and was able to fly it for a few seconds until it suddenly went into a nose dive again. I cycled S1 and S2 again, left S1 in ATTI and gave it full throttle again. I recovered about 5ft above ground. (I'm pretty sure S2 was moved back into OFF.) I flew back in ATTI fine but it was extremely windy with gusts so my son grabbed it when it was in a relatively stable hover.

Every event happened in a couple of seconds and although I'm pretty sure what my recovery control inputs were I'm not 100% sure. Things happened so fast.

I follow a check list and prior to heading out for the flight, I keep the P2V+ close so I can listen to its sound and run through all the controls to make sure everything is working. I had 12 satellites, all batteries were fully charged. The K-index was 3 at the time. The flight started off completely normal. I was about 10 minutes into the flight.

Now, I realize that the winds were likely too strong to fly and maybe down drafts and wind shear caused some problems but what threw me was the way my P2V+ buggered off from a hover. And entered into a vertical dive twice. Talk about a WTF moment. I've had about a 100 flights and always had the “fly away” problem in the back of my mind so I was ready for it in a way. Stunned but aware. (I'm not a complete noob to aviation as I used to fly for a living.)

Was this a fly-away or just some really weird local weather?

The question bugging me is if that happened to me, whether I would be able to ever trust it again. Or whether I would consider swapping out the naza first. If it is not a hardware issue (e.g. faulty IMU component) there might be little point doing this.
 
If you want to be absolutely certain you will not get into a flyaway situation you should consider going back to U-Control flying models (for those of you too young to know what that is check this out http://en.m.wikipedia.org/wiki/Control_line)
Servos crap out, batteries fail, hydraulics leak, gas lines crack, motors seize and in general s!?'t happens. I simply think it is remarkable tha DJI and others have created something as reliable as this has been. Yes it hurts financially when you auger in or watch it go away but you signed up for the hobby. Let's pull up our big boy pants and stop complaining about how screwed up DJI is. BTW for the Walkera fans I suggest you do a quick search of their forums to find the same complaints about flyaway and sudden course anomalies, some of which complain that DJI does a better job than Walkera on quality control. I haven't lost one yet but I've come close. Being a 21st century American, my first reaction was to believe someone else is to blame (DJI, FAA, CIA), my second reaction was to decide if I was willing to risk another flight or sell the contraption. So far I have taken the risk because the fun factor outweighs the potential costs.
 
I had some weirdness yesterday. Was the second time it happened. Went to a park with the P2H4 to do some control exercises. First battery was fine, the drone flew consistent and predictable.. Second battery, when I took off in GPS mode (With 11 Sats) the drone was pitching foreword with a centered collective. I compensated to maintain a hover and over the course of maybe 15 seconds it slowly corrected itself and found it's neutral stability.. The rest of the flight was fine, regardless of ATT or GPS mode.
 
I know this may sound too much like “grassy knoll” but has anybody considered an outsider intentionally hacking into Phantoms?

Anything can be hacked and I seriously doubt these things are “locked down” to the point they’re 100% immune from a talented and determined individual looking to mess with a flyer.

I’m also sure there’s plenty of legitimate technical reasons Phantoms fly-away but I wouldn’t rule out an intentional act either.
 
What's the goal of hacking a phantom?

What's the goal of hacking anything? Maybe just to see if it can be done? Maybe somebody banned from a discussion forum and carrying a grudge? Maybe those people who threaten to "blow it outta the sky"? Maybe just some kids messing with us?

Plenty of sick yet intelligent people out there. I'm not an advanced tech but I'm fairly certain an RF data logger could reveal a ton info about ways to hack these birds.
 
an unexperienced linux USER (not even a hacker.. lol just a user) could do damage when they 'hacked' in... not take control, but permanantly stuff up the wifi module or reboot it...

wont go into details, but its basic stuff... SSH, default root passwd.. and problem is if you change the roof password some portions of the DJI app may not work.. but someone can easily SSH into a Phantom and yeah either kill the wifi module so it needs to be sent back to dji or keep rebooting it over and over.. or setup a script to reboot on bootup, everytime, again needing it to be sent back.
Same concept applies to the Wifi module, RE and Camera as they can all be accessed via SSH.

Very easy fix for DJI to implement to stop that problem without restricting access to SSH or functionality.... They just need you to be able to let you change root passwd (which is easy enough at the CLI) but make it still work 100% with the app... so perhaps make it so if you change the root password you can somehow tell the app what the new password is.

I've been reading about spoofing GPS data soon and its funny reading some of the comments (not on this forum but others) everyone is like "oh yeah that would be so easy to do..i just cbf testing but yeah so so easy"... yeah yeah sure.. perhaps for some, but not for majority.


Anyways.........
I had a flyaway. Was a while ago so trying to remember exact details.. but had used both batteries maybe 50 charges each... flew at the same park 80% of the time.. was on the second battery for the day... 5 mins in... phantom was 200ft horizontal, 100ft veritcal... flying along horizontally.. it stops suddenly... hovers.. stops reponsinding to RC inputs... then continues on its original horizontal course I was flying it on, UPWIND mind you, plus there was minimal wind anyways.... had no RC control what so ever, no sticks worked, no switches.. it was like RC was not there... luckily after 10-15 seconds it initiated an RTH but its self as it relised the RC was not connected anymore and landed safely, well hand caught it on that occasion just to be sure.

Anyways once on the ground it still didn't work.. rebooted RC, no luck.. only thing to fix is was rebooting the phantom.. then it has been perfect/never happened again for another 100+ charges of each battery, still hasnt happened again *touch wood*

originally I thought it was must of been radio interferance as there were some commercial antennas not to far away I guess, and 5.8ghz is popular for some commercial wifi systems but I figured if it was interferance surely once out of the path it would resume normal operation, but perhaps the amount of interfance was so great it pissed off the NAZA/FC/IMU/Whatever until a reboot cleared it..

took a while to get confidence back...

but yeah..good times :)

*edit* had my vertical/horizontal mixed up
 
Last edited:
  • Like
Reactions: Jeff48920
Actually just on the hacking thing... after 3 mins more thought... I suppose someone with basic 'hacking' tricks could take control of someones app/ground station and take control of the flight.

Not going to go into the details of how or why, just so people don't try it... (allthough PM me if you are familiar with linux and want to learn about this stuff and not actually do it and I'll tell you my specifics)

but if user X is connected via the app just like normal... As we all know an unauthorized person (or friend) on the same wifi network can't connect to the app the second time(at the same time as user X)... However it would be very easy to kill the connection of the app from user X to the Phantom, kill his IP so they can't reconnect but keep trying... and while that goes on the 'hacker' connects to the app and sets it to go on G/S mission... obviously user X still has ultimate control, as they have the RC and just needs to change to ATTI, but it could all happen so fast they have no time to switch modes... Could do same trick while its flying in the air (in GPS mode) or on ground powered up (in GPS mode)

I'd only be worried of this scenario, and the scenario I mentioned in the first post if you go to a Phantom training or something with LOTS of phantom users, and some d*ck head wants to cause trouble.
But in summary, would be very easy to do all i've mentioned. And probably not the first time its been mentioned lol... everyone slightly tech has know about all these potential issues since day 1.
 
It's got a bit of a learning curve to it but all the information is there. Pawelsky is the mastermind who figured it all out. And he might be making more of the boards.

http://www.rcgroups.com/forums/showthread.php?t=2071772

Have you looked into this at all ??
https://github.com/noahwilliamsson/dji-phantom-vision/blob/master/dji-phantom.c
https://github.com/noahwilliamsson/dji-phantom-vision

He gets the data from the app and phantom (well wifi module) but its fed data from the gps and??? naza??
Seems like its the data in/out... almost 3am here so details are abit fuzzy :/

You don't need any additional hardware to do what he is doing, just some linux skills..

I've been being looking at that for a little while, but haven't seen the thread you posted up until coincidentely a few hours ago.. Haven't gone through the hole thread yet seems its so big, but figured I'd mention it seems from what I read so far it sounds like the 2 are quite similar ?

Let me know if you are familiar with the link I just posted and know if its the same or diff etc etc ?

Might read your thread tomorrow.. just need to find a good few hour block as those long tech threads take a while to read seems you have to read a few posts multiple times to understand it.. lol, well atleast I do :(
 
What's the goal of hacking anything? Maybe just to see if it can be done? Maybe somebody banned from a discussion forum and carrying a grudge? Maybe those people who threaten to "blow it outta the sky"? Maybe just some kids messing with us?

Plenty of sick yet intelligent people out there. I'm not an advanced tech but I'm fairly certain an RF data logger could reveal a ton info about ways to hack these birds.
They would need lots of time on their hands. Considering people don't regularly fly drones in the same place all the time.. The hacker would need to drive around looking for flying phantoms.
 
Pretty interesting but that is Vision only. It has less data than the CAN bus implementation but it can send commands. CAN implementation commands are not done yet. I would really like to be able to control the NAZA via CAN. It's possible but not implemented yet.
 
They would need lots of time on their hands. Considering people don't regularly fly drones in the same place all the time.. The hacker would need to drive around looking for flying phantoms.

Missed my disclaimer - "grassy knoll", etc?

But if/when I suffer a fly-away be assured I'll 1st break out the binocs to scan the area very carefully for "suspicious" activities/individuals. Then go look for the bird...
 
Hacking doesn't have to be a bad thing, depends on your intention. Reading data from the CAN bus can already be considered hacking because it isn't something the manufacturer thought users would do. Most mods here like the GPS puck are also hacking.
Hacking something to intentionally cause harm is a different thing. Someone could reverse engineer the software, offer hacked firmware for download with "malware", take over your bird. Depends if the reward would be interesting enough to justify all the time put into reverse engineering everything.
If you take a look at www.hackaday.com, you'll see that there's tons of good hacking going on.
 
  • Like
Reactions: Ti22
I’m sorry for using that “H” word. It’s much like the “D” word - too much negative connotation. I’m old, I use it to say “people I dunno doing things to my stuff”. Some hacking is good, I agree. Some is bad, I agree too. No offense, sorry if it did…
 
None taken. Its just public ignorance just like in the case of drones. Hackers helped shape many of today's technologies, and some like Mitnick were just used as exemplary "bad guys" to scare people from being curious. The word 'hacker' was the 90's equivalent of 'terrorist' today, in that it is intentionally misused for fear.
 
  • Like
Reactions: Ti22
Unbelievably, it happened AGAIN!

12 minutes and 45 seconds into last night's first flight, the Phantom lost it's ability to stay level and plummeted to the pavement at roughly 15m/s. It had power all the way down. I was very lucky it didn't land on an adjacent roof top. Damage is worse than last time. This Phantom is dead.

Unlike last time, I have a black box full of data right up to the impact. It's too early to tell, but it looks like the NAZA got some very strange data. At one point, it reported I had 204 GPS satellites! I am dissecting it now and looking for the smoking gun. Video, unlike last time, I didn't let the camera repair the file. I am looking for ways to recover the file entirely so I do not lose the crash.

Stay tuned for a full Air Crash Investigation. DJI will be paying for the replacement (they just don't know it yet). ;)
 
  • Like
Reactions: RichWest and Ti22
Talk about drunken sparrow, this is a druken suicidal sparrow.
Really interested to see the data and find out what causes these incidents. Sorry for your loss, Ian.
Also wondering if you tried flipping into Naza mode or there wasn't enough time for that.
 
Unbelievably, it happened AGAIN!

12 minutes and 45 seconds into last night's first flight, the Phantom lost it's ability to stay level and plummeted to the pavement at roughly 15m/s. It had power all the way down. I was very lucky it didn't land on an adjacent roof top. Damage is worse than last time. This Phantom is dead.

Unlike last time, I have a black box full of data right up to the impact. It's too early to tell, but it looks like the NAZA got some very strange data. At one point, it reported I had 204 GPS satellites! I am dissecting it now and looking for the smoking gun. Video, unlike last time, I didn't let the camera repair the file. I am looking for ways to recover the file entirely so I do not lose the crash.

Stay tuned for a full Air Crash Investigation. DJI will be paying for the replacement (they just don't know it yet). ;)

Terrible news Ian.
:(

At least you might have the answer to DJI Phantom flyaways in your grasp.

Edit. I wonder if it possible to make a sectory level duplicate of the SD Card, before the FAT is modified ?
 

Recent Posts

Members online

Forum statistics

Threads
143,094
Messages
1,467,597
Members
104,980
Latest member
ozmtl