Air 2S darting forward and crashing into a roof twice!

[BigAl07]

I took off from the roof of my car some 100 feet away as the road was on a slope and this was the most level option. I didn't notice home point being updated during flight.

Welcome to the forum Johny. I'm Allen, the guy who suggested you join here and have your flight logs analyzed.

I can tell you from decades of experience both flying and instructing, there are times when we "move the sticks" and in our minds we are 100% positive it's not how it happened. If the Flight Log indicates stick movement, what-so-ever, you can pretty much take that to the bank as gospel. Maybe you didn't intend or even realize it but somehow the stick was moved. In Sport Mode, close to several obstacles, and probably a bit nervous is a recipe for "bent/broken" drones for all but the best of us.

Hopefully you get it flying again and do some practicing in Sport Mode well clear of any obstacles.

Safe Flights,
Allen

 

[JohnySVK]

Welcome to the forum Johny. I'm Allen, the guy who suggested you join here and have your flight logs analyzed.

I can tell you from decades of experience both flying and instructing, there are times when we "move the sticks" and in our minds we are 100% positive it's not how it happened. If the Flight Log indicates stick movement, what-so-ever, you can pretty much take that to the bank as gospel. Maybe you didn't intend or even realize it but somehow the stick was moved. In Sport Mode, close to several obstacles, and probably a bit nervous is a recipe for "bent/broken" drones for all but the best of us.

Hopefully you get it flying again and do some practicing in Sport Mode well clear of any obstacles.

Safe Flights,
Allen

Allen, thank you for introducing me to this forum. I have learned how to pull the flight log and it certainly provides some interesting data!

I understand the skepticism on this forum regarding my flight skills especially since the data seems to be against me, but I am adamant about my statement. There is absolutely no chance I would push the stick full forward in this situation. I use the two-finger technique and can assure you this was not a pilot error. The reason I am confident about it is the fact that it happened twice in a row under the exact same circumstances. I will be sending my DJI in for the diagnostics and will post the results.

Thanks again for your help.

 

[mobilehomer]

Again, check your gain and expo settings.

 

[Meta4]

I am adamant about my statement. There is absolutely no chance I would push the stick full forward in this situation. I use the two-finger technique and can assure you this was not a pilot error.
The reason I am confident about it is the fact that it happened twice in a row under the exact same circumstances.

I guess the flight data, just got it wrong for 10 seconds or so for the first impact ... and again 30 seconds later for the second.
But it's strange how the drone's pitch, speed increasing and crash impacts seem to match your description of what happened so well.
It's really strange that the joystick inputs fit in perfectly well with the other data.

Under stress, what people think happened isn't always what actually happened.
Recorded flight data doesn't get confused and make up things to describe a flight.

I will be sending my DJI in for the diagnostics and will post the results.

Good luck with that.
DJI rarely give any detailed explanations.

 

[JohnySVK]

I guess the flight data, just got it wrong for 10 seconds or so for the first impact ... and again 30 seconds later for the second.
But it's strange how the drone's pitch, speed increasing and crash impacts seem to match your description of what happened so well.
It's really strange that the joystick inputs fit in perfectly well with the other data.

Under stress, what people think happened isn't always what actually happened.
Recorded flight data doesn't get confused and make up things to describe a flight.

Not necessarily, I'm sure the data provided from flight log are correct because that's exactly what happened - the drone was hovering, then received a signal to go full forward and it did, but can you claim with 100% certainty this input came from the controller I was holding? That there is no chance there might have been some kind of interference? I guess all the flyaway accidents that happen are just pilots under stress holding the controller full forward. I flew in tight spots before, I flew missions where the risk of damaging the drone or property were much higher than that day and my fingers never slipped. This was hardly a stressful situation. I'd also like to point out that there was no attempt to stop the drone once it took off. Anyone who sees their drone heading for an obstacle will instinctively pull back and try to stop it. I guess I was just watching my drone for 10 seconds darting towards the roof and I didn't think to prevent the accident. Twice...

Good luck with that.
DJI rarely give any detailed explanations.

At this point I don't see what other options I have. Unless I go back and repeat the mission risking third accident and loss of drone just to prove that I was right? Not worth it.


That being said, I still do appreciate the input of everyone here and thanks to BigAl07 for introducing me to this forum.

 

[Meta4]

Not necessarily, I'm sure the data provided from flight log are correct because that's exactly what happened - the drone was hovering, then received a signal to go full forward and it did, but can you claim with 100% certainty this input came from the controller I was holding?

Yes

That there is no chance there might have been some kind of interference?

You are clutching at straws.
Interference doesn't send precise joystick input signals to drones.

I guess all the flyaway accidents that happen are just pilots under stress holding the controller full forward.

Analysis of flight data from hundreds of incidents reported as "flyaways" has shown that almost all are cases of operator confusion or operator error.

I flew in tight spots before, I flew missions where the risk of damaging the drone or property were much higher than that day and my fingers never slipped. This was hardly a stressful situation. I'd also like to point out that there was no attempt to stop the drone once it took off. Anyone who sees their drone heading for an obstacle will instinctively pull back and try to stop it. I guess I was just watching my drone for 10 seconds darting towards the roof and I didn't think to prevent the accident. Twice...

I can't tell what you are trying to say here, but the data unequivocally shows exactly what happened and it shows it very clearly twice.

At this point I don't see what other options I have.

What options?
One possible option might be to shake of the denial and accept what the flight data shows.
That would be a lot easier than trying to explain how the recorded data is lying.

Unless I go back and repeat the mission risking third accident and loss of drone just to prove that I was right? Not worth it.

It wouldn't be worth it because it wouldn't prove anything.

 

[Phantomrain.org]

Ok, so here is my take on this :

First , there is some pilot errors:

Taking off without getting the Home Point
Flying in sport mode
Risking Magnetic Interference with the Compass by taking off from the roof without a Home Point.

Capturing the Home Point 40 ft in the air , might have put your drone in terrible position if it would have gone into RTH at any time in the flight. If that happens again you need to know you can reset the homepoint while in the air to a safer postion. You could also land on the roof to reset the homepoint as well if possible. Bare in mind if the angle is to steep than you will not be able to take off again once the motors shut down.

As someone who flys off of my truck / Car/ Bike because of a bad back I would repeat the Flight off the Roof of the car to see if that in fact influenced the flight in a safe area, also check to see how long it took to get the home point.

I do have to add , that the Air 2S has not shown any Magnetic Interference warnings in any of my Test Flights thus far but I have had a few Compass Errors that caused me to have to restart off the hood,


When it comes to the Data : Its not the 100 % Tell all to DJI , so sending in a Claim in this case is reasonable. There going to ask you for Video or even the Cashe Video to help them determine the claim. Good luck,


Phantomrain.org
Gear to fly in the Rain. Capture the Storm and Fly from you car.

 

[Meta4]

First , there is some pilot errors:

Taking off without getting the Home Point
Flying in sport mode
Risking Magnetic Interference with the Compass by taking off from the roof without a Home Point.

Capturing the Home Point 40 ft in the air , might have put your drone in terrible position if it would have gone into RTH at any time in the flight. If that happens again you need to know you can reset the homepoint while in the air to a safer postion. You could also land on the roof to reset the homepoint as well if possible. Bare in mind if the angle is to steep than you will not be able to take off again once the motors shut down.

As someone who flys off of my truck / Car/ Bike because of a bad back I would repeat the Flight off the Roof of the car to see if that in fact influenced the flight in a safe area, also check to see how long it took to get the home point.

I do have to add , that the Air 2S has not shown any Magnetic Interference warnings in any of my Test Flights thus far but I have had a few Compass Errors that caused me to have to restart off the hood,

None of this had any effect on the flight.
What caused the incident has been explained in posts #4 and #9.

When it comes to the Data : Its not the 100 % Tell all to DJI , so sending in a Claim in this case is reasonable. There going to ask you for Video or even the Cashe Video to help them determine the claim.

It sounds like the drone was not damaged so there's no claim to be made.

 

[rjwmorrell]

Not necessarily, I'm sure the data provided from flight log are correct because that's exactly what happened - the drone was hovering, then received a signal to go full forward and it did, but can you claim with 100% certainty this input came from the controller I was holding? That there is no chance there might have been some kind of interference? I guess all the flyaway accidents that happen are just pilots under stress holding the controller full forward. I flew in tight spots before, I flew missions where the risk of damaging the drone or property were much higher than that day and my fingers never slipped. This was hardly a stressful situation. I'd also like to point out that there was no attempt to stop the drone once it took off. Anyone who sees their drone heading for an obstacle will instinctively pull back and try to stop it. I guess I was just watching my drone for 10 seconds darting towards the roof and I didn't think to prevent the accident. Twice...

At this point I don't see what other options I have. Unless I go back and repeat the mission risking third accident and loss of drone just to prove that I was right? Not worth it.


That being said, I still do appreciate the input of everyone here and thanks to BigAl07 for introducing me to this forum.

Fly away incidents are rare, and completely different than commanded flight movement (whether it was commanded by you or something else, it's all the same to the drone, the log indicates a forward command - I use stick mode 3 as well).

Fly away is rare in GNSS drones because the onboard flight systems rarely fail that badly. The drone knows where it is.

 

[rjwmorrell]

I have not analysed dozens of logs or incidents, but there does appear to be a pattern of pilot error here and some misunderstanding about the flight modes and capabilities of the aircraft.

The forward stick movement is clear in the logs. Whether the pilot commanded those inputs or not isn't really relevant, it's clearly recorded by the flight log and it only happens during the sequences where the pilot reports the incidents.

Flying in sport mode is a known risk factor for collisions. The aircraft has faster response, higher speeds, and no collision avoidance systems active in sport mode. Operating in sport mode at low altitude in proximity to buildings and trees is a compoundingly high risk factor.

While not causes of the incident, potential indicators of inexperience or inattention include taking off without full satellite connections and the updating of the home point 7m/20ft from the launch location, as well as no function test after launch. Another factor I would consider is the low altitude, sport mode flight and distance from the launch site - and potentially the pilot. If the pilot stayed stationary at least the launch site for the entire length flight, it is hard to see how they could have had adequate situational awareness at the low altitudes (sometimes as low as 4 feet VPS) while maneuvering.

DJI recommends a minimum distance from buildings and obstacles, and operating within that minimum certainly increases the risk of collision, radio interference, loss of situational awareness, and reduces reaction time during flight, especially sport mode.

Under pressure of imminent or occurring collision, pilots sometimes react in unpredictable ways or default to muscle memory, formerly learned behaviours etc. Without knowing how often this pilot flies in stick mode 3, it's entirely possible they intended to command an increase in altitude instead of a forward movement by accident.

A screen capture or video of the pilot using the controls could help determine that, but that doesn't appear to be available.

 

[JohnySVK]

Ok, so here is my take on this :

First , there is some pilot errors:

Taking off without getting the Home Point
Flying in sport mode
Risking Magnetic Interference with the Compass by taking off from the roof without a Home Point.

Capturing the Home Point 40 ft in the air , might have put your drone in terrible position if it would have gone into RTH at any time in the flight. If that happens again you need to know you can reset the homepoint while in the air to a safer postion. You could also land on the roof to reset the homepoint as well if possible. Bare in mind if the angle is to steep than you will not be able to take off again once the motors shut down.

I never really paid attention to the Home Point update before but I'm starting to understand the importance. I don't think this situation had anything to do with HP, because the satellite signal was strong and I was 10 feet from the drone as I followed it while navigating around the trees. Even if the drone decided it wants to engage RTH, the default move is up. Now I am aware there are situations when the drone will not go up before heading the direction of the HP, but in this case the HP was in the other direction and the few times I tried RTH, the drone never took off 10 mph in the direction of the HP.

As someone who flys off of my truck / Car/ Bike because of a bad back I would repeat the Flight off the Roof of the car to see if that in fact influenced the flight in a safe area, also check to see how long it took to get the home point.

I took off from the roof of my car at least 80 times and I never had any issues.

I do have to add , that the Air 2S has not shown any Magnetic Interference warnings in any of my Test Flights thus far but I have had a few Compass Errors that caused me to have to restart off the hood,


When it comes to the Data : Its not the 100 % Tell all to DJI , so sending in a Claim in this case is reasonable. There going to ask you for Video or even the Cashe Video to help them determine the claim. Good luck,


Phantomrain.org
Gear to fly in the Rain. Capture the Storm and Fly from you car.

I don't have a video as I was in a photo mode, and I don't know how to retrieve the cash video. I just know I have nothing to lose even if they tell me the same thing I've heard here. That I am a lousy pilot lol.

 

[JohnySVK]

I have not analysed dozens of logs or incidents, but there does appear to be a pattern of pilot error here and some misunderstanding about the flight modes and capabilities of the aircraft.

The forward stick movement is clear in the logs. Whether the pilot commanded those inputs or not isn't really relevant, it's clearly recorded by the flight log and it only happens during the sequences where the pilot reports the incidents.

Flying in sport mode is a known risk factor for collisions. The aircraft has faster response, higher speeds, and no collision avoidance systems active in sport mode. Operating in sport mode at low altitude in proximity to buildings and trees is a compoundingly high risk factor.

While not causes of the incident, potential indicators of inexperience or inattention include taking off without full satellite connections and the updating of the home point 7m/20ft from the launch location, as well as no function test after launch. Another factor I would consider is the low altitude, sport mode flight and distance from the launch site - and potentially the pilot. If the pilot stayed stationary at least the launch site for the entire length flight, it is hard to see how they could have had adequate situational awareness at the low altitudes (sometimes as low as 4 feet VPS) while maneuvering.

DJI recommends a minimum distance from buildings and obstacles, and operating within that minimum certainly increases the risk of collision, radio interference, loss of situational awareness, and reduces reaction time during flight, especially sport mode.

Under pressure of imminent or occurring collision, pilots sometimes react in unpredictable ways or default to muscle memory, formerly learned behaviours etc. Without knowing how often this pilot flies in stick mode 3, it's entirely possible they intended to command an increase in altitude instead of a forward movement by accident.

A screen capture or video of the pilot using the controls could help determine that, but that doesn't appear to be available.

Thank you for your response. I was following the drone as I was navigating it around the trees, so I was some 10 feet from it when the incidents occurred maintaining VLOS the entire time.

I've tried to use the default stick mode when I purchased the drone, but after some 10 flights I gave up and switched to stick mode 3 as it felt much more familiar. I'd also like to point out that when clearing obstacles like that I used the 'stair pattern.' I move the drone in an L shaped increments in the desired direction instead moving both sticks at the same time. The pattern is clearly visible on my flight log, which means the stick move at 4.4.1 (4.31.1 respectively) makes no sense because that's when I'd have returned the stick to the neutral position. But clearly that's not what happened.

 

[rjwmorrell]

Thank you for your response. I was following the drone as I was navigating it around the trees, so I was some 10 feet from it when the incidents occurred maintaining VLOS the entire time.

I've tried to use the default stick mode when I purchased the drone, but after some 10 flights I gave up and switched to stick mode 3 as it felt much more familiar. I'd also like to point out that when clearing obstacles like that I used the 'stair pattern.' I move the drone in an L shaped increments in the desired direction instead moving both sticks at the same time. The pattern is clearly visible on my flight log, which means the stick move at 4.4.1 (4.31.1 respectively) makes no sense because that's when I'd have returned the stick to the neutral position. But clearly that's not what happened.

I'm glad to hear you were following it, that clears that up. I agree the default stick mode (Mode 2) is not intuitive. I play video games too, Mode 3 is what I use.

By the way, now that you know you can manually turn off Obstacle Avoidance, in the future consider flying these missions in "Cine" flight mode. You can even manually adjust the responsiveness for each flight mode, so you can manually set your Cine mode to be even slower and less jerky/responsive in the settings within DJI Fly. It's something I learned on a YouTuber's webinar on getting the best cinematic settings out of the DJI products. It also allows you to fly more safely when near to obstacles and objects, because you have a bit more time to react if the aircraft starts doing uncommanded or unexpected things. One final suggestion -- you're over 250g anyway with an Air 2S, so consider prop guards if you are doing flight close to objects. It will not only save the props if there is a collision, but save an injury if you're in proximity to people.

 

[JohnySVK]

I'm glad to hear you were following it, that clears that up. I agree the default stick mode (Mode 2) is not intuitive. I play video games too, Mode 3 is what I use.

By the way, now that you know you can manually turn off Obstacle Avoidance, in the future consider flying these missions in "Cine" flight mode. You can even manually adjust the responsiveness for each flight mode, so you can manually set your Cine mode to be even slower and less jerky/responsive in the settings within DJI Fly. It's something I learned on a YouTuber's webinar on getting the best cinematic settings out of the DJI products. It also allows you to fly more safely when near to obstacles and objects, because you have a bit more time to react if the aircraft starts doing uncommanded or unexpected things. One final suggestion -- you're over 250g anyway with an Air 2S, so consider prop guards if you are doing flight close to objects. It will not only save the props if there is a collision, but save an injury if you're in proximity to people.

That's a good idea. Thanks again for the suggestions!