Consistent GS flyaway at first waypoint with severe consequences

Com

Joined
May 22, 2015
Messages
10
Reaction score
0
Age
35
I have flown Phantom 1, 3DR drones and my own home-built pixhawk/APM quadcopters for 3 years with autonomous flight with no flyaway issues. This is my first time flying with GS on Phantom 2. I was doing it as part of a comparison among autonomous consumer-grade drones as part of a review article.

I have a brand new, stock Phantom 2 + Ground Station (PC) bought directly from the DJI website with no other payload than the data link module. The antennas were positioned one horizontal and vertical, down the leg with no other antenna or compass on it, per instructions. The firmware was updated to latest. All software was downloaded from the DJI website this week. It's flown in Phantom mode. Flies nice and stable on RC. When I let go of controls it hovers in its place even in 15 mph winds for over 10 minutes. GS connects and records tracks and reports flight telemetry fine.

When I tried to execute a mission at consisting of 6 waypoints 20-80m apart, it flew to the first waypoint without incident, and then suddenly veered away in a wide arc until it crashed sideways at full speed.

No input from ground station or the remote control could snap it out of it. Tried toggling the S1/U switch (the one on the right with the GPS/ATTI/ATTI markings) to regain TX control but did not work. RTL failsafe was configured as the third position of the switch but also did not work.

Battery thrown clear. Props broken. One line of lights busted.


Maybe it was an isolated incident. I opened the case to check for any internal damage, tightened all the screws nice and tight and checked that the motors were clear of the shrouds, replaced all the props (I have like 3 full sets of spares), grabbed a fresh battery, checked all the configurations and double-checked. I re-read all the manuals cover to cover. I reinstalled all the software and went through every option. I recalibrated with the assistant, both basic and advancd, and also recalibrated the compass/gyro outside. I did some flight tests without waypoints. All good. Solid hover, turning, movement, ascent/descent. It knows where it's facing, how high it is, nice and level, full GPS lock. No wind, nice balmy 24C.

I give it another simple set of waypoints 10-50m apart, this time specify 3m/s speed, stop and turn with 3 second wait, 12m altitude so I can keep a closer eye on what's going on. Upload. Upload success. Nobody in the field (a private property soccer field, but slightly longer) or anywhere near the flying area. Arm, liftoff, watch for any erratic behaviour. Nope, rock solid. Should be perfect. Hit the go button.

Again it makes a beeline for the first waypoint. Reaches first waypoint at one corner of the field. And then without waiting it shoots away at extreme speed in a wide, gradually descending arc - think a very forceful frisbee curve throw, but longer, faster and higher - same way as before. I realized it could be on a collision course with someone walking by outside the opposite corner of the field where I was flying. I tried aborting the mission through GS. I tried toggling the flight mode switch. I tried RTL. I even tried to disarm - better it crash where it was than hit the person.

No response to either GS or TX even though GS was still connected and tracing the flight path and the instrument panel showed the phantom knew the strange way it was flying. I watch helplessly as it flies at the unaware passer-by like a missile. It nearly shreds the person's face and chest, except they made a last second dodge and the spinning blades missed by centimetres. The phantom crashes into the sidewalk outside the field. The battery is flung out, the props are broken, the frame is slightly warped. Suffice to say I was in shock and had trouble speaking to this person, who was understandably beside himself with anger. The incident was reported to the authorities.

Funny enough my flights were being done as part of a bigger review of the autonomous capacities of consumer-grade drones and DJI does not exactly instill confidence with this performance. First, there is the sudden catastrophic flyaway issue at hand. Second, there is the total loss of control by either means even with a solid connection. Third, the failsafes failed. Fourth, this happened both times. And I am not going to risk a third.

The close call was a freak incident, but someone could have been maimed for life or even killed because of this glitch. Spinning props lacerating someone's jugular approaching from the sky at over 40mph? No time to react, no contest. This is why people want drone laws to be in place, and have it regulated so that only flights in controlled areas with safety buffers are allowed. I'm kind of in their camp now even though it would make things harder for myself.

But I was wondering if anybody had any ideas why this happened. The circumstances around the flyaway and crash - which happened in the same way both times - as well as the flawless non-waypoint and first-waypoint flight performance leaves very little for me to identify as a cause. My only conclusion thus far is of a dangerous glitch that triggers on arriving at the first waypoint. But why hasn't it happened to others?
 
Last edited:
Try reposting in the P2 forum.

You'll get more eyeballs there.
 
Try reposting in the P2 forum.

You'll get more eyeballs there.

This is a GS issue though, and I clicked on the Ground Station subforum of the P2 forum and got here. The P2 sans Ground Station seems to fly flawlessly.

Good to have more eyeballs though, I might after a while.
 
Sure,

I see your point.

Good Luck to you.
 
Just so you are better informed there is no ATTI in the phantom mode. The only time you would have ATTI in the phantom mode is if you had gone below 6 satellites or lost GPS altogether.
 
Just so you are better informed there is no ATTI in the phantom mode. The only time you would have ATTI in the phantom mode is if you had gone below 6 satellites or lost GPS altogether.

Yes; I was only referencing the switch on the RC TX controller, which is marked in such a way, not the flight mode itself. Supposedly toggling this switch down and up should return control from autopilot to the RC TX. It didn't work. Neither did the third position which I set to RTL in the assistant software.
 
In order to have ATTI in the center position of S1 you have to switch from phantom to naza-m mode but you will not have IOC unless you turn that function on. So when in naza-m S1 is GPS, ATTI and Failsafe if you set it up in the software.
 
Just so you are better informed there is no ATTI in the phantom mode. The only time you would have ATTI in the phantom mode is if you had gone below 6 satellites or lost GPS altogether.
If I recall correctly, you can't get into GS from Phantom mode either. I might not remember this correctly.
 
So you were initially running GS without any video? Okay.... but if you knew there was an issue and were going to do the same thing again why would you not screen record, film the flight or turn the video on in the Phantom?
 
Can't help you with GS. Sorry.
 
In order to have ATTI in the center position of S1 you have to switch from phantom to naza-m mode but you will not have IOC unless you turn that function on. So when in naza-m S1 is GPS, ATTI and Failsafe if you set it up in the software.

I kept it in phantom mode and had no intention of using ATTI. Again, I was only referring to the markings on the TX beside the S1 switch. I'll edit my post to reduce confusion.
 
I know what the switch marking say and that can be confusing for someone new to the hobby.
 
So you were initially running GS without any video? Okay.... but if you knew there was an issue and were going to do the same thing again why would you not screen record, film the flight or turn the video on in the Phantom?

I have the base Phantom 2, there is no camera mounted sadly. (Intention was to mount my own APS-C camera later). And honestly on the ground I have my hands full with GS and the TX. It would have been nice to have someone else to take a video though to show what happened.
(Except if it killed someone, which it nearly did, that could be evidence against me in court for manslaughter...)
 
Last edited:
If I recall correctly, you can't get into GS from Phantom mode either. I might not remember this correctly.

Yeah, I couldn't find a way to activate the mission from the TX. It must be done by hitting the GO button in the GS software it seems - or is there another way?
 
I kept it in phantom mode and had no intention of using ATTI. Again, I was only referring to the markings on the TX beside the S1 switch. I'll edit my post to reduce confusion.


Had you set the MODE to NAZA vs Phantom you would have been able to regain flight control by the S1 switch
GPS to ATTI to GPS

if you are in phantom mode this will not function
 
Had you set the MODE to NAZA vs Phantom you would have been able to regain flight control by the S1 switch
GPS to ATTI to GPS

if you are in phantom mode this will not function

Ah that would explain why one of the things I tried to regain control with didn't work. I didn't see this in the GS manual. It should be like frickin front and centre. So with GS in the default Phantom mode is there no way to ever switch back to TX?
 
Sorry, I thought this was a newer model.

But that's a good thought if I ever try to test this thing again... far away from any buildings or people... still have 2 sets of spare propellers after all... I should make sure to have a second person to record and maybe a screen capture running. At least with that maybe I can get something back from DJI.

Do you know of any good free screen capture software?
 
that would not be involved with GS , That is main feature of the quad IOC and failsafe all need NAZA mode enabled

I don't care about IOC and in Phantom mode failsafe can be enabled on S1 position 3 in assistant, which I did. The failsafe did not work during the flyaway. I'm interested in how to get control back from a GS-initiated mission, and why my attempts didn't work. Aborting a mission to TX control seems a basic requirement for autonomous flight but in this case it was all FUBAR.
 

Recent Posts

Members online

Forum statistics

Threads
143,066
Messages
1,467,357
Members
104,935
Latest member
Pauos31