I have an Inspire 1 Pro with just under 20 hours on it. This week it smashed into the floor and destroyed itself. Clearly, DJI's response was simply "out of warranty, sorry".

Here is what happened:  I was filming on a dry lake bed. At the end of the shoot, with about 50% battery remaining, I turned the aircraft back towards the home point at about 40ft and hit the max speed forward control. DJI playback shows that I didn't touch the left stick (up-down).  For the first 15 secs, the drone maintained level and closed from 900ft to 500ft range. At this point, I took my eyes off the screen to identify the drone in the air. 4 seconds later there was a smash as it impacted the dust at 35 mph.

Playback shows that around the 500ft range to home point, the drone started about a 10-degree dive until it hit the floor. The GPS heights showed a slight descent from 40ft to an impact height at 26ft despite it hitting the floor. There was no apparent attempt for the system to attempt to regain 40ft. After impact, the GPS height continued to descent to about -500ft before slowly returning to +26ft.

DJI have offered to charge me to fix it - wow, too kind! My question is, what is the point? If the POS now has software that is incapable of height-keeping despite having 14 active satellites and 5 bars of signal strength in a remote RF interference-free environment, then I can certainly not trust it around any sort of built-up area or in the vicinity of people. It could cause considerable damage and potentially kill someone.

No doubt someone here will usefully chime in with, "yeah, old tech, worn out". Really? A piece of cutting edge tech that worked perfectly well is now not expected to operate as before because it has 20 hours on the thing?  The battery had approx 4 cycles on it and was showing normal power output.

Personally, it wouldn't surprise me if the software "update" I completed 3 days prior, having not used the thing for over a year, was designed to be unstable so that these things are all eventually removed from service while blaming pilot error.

I will try and work out how to upload the logs so that you guys can all take a look and decide for yourselves how safe these things are to use. In the meantime, has anyone else experienced the same issue?

Regards,

What FW version were you at before the upgrade?
There have been changes over time in FW that had major effect to how the Inspire flies.

All I know is that it is now 1.11.01.50+ on the aircraft and 1.7.80 on the controller. Every time I use it I have to make sure that everything is up to date because I once had an iPad IOS that wouldn't talk to the current ac FW and ended up flying the whole day using my iPhone which had a different IOS.  Here is the Airdata log if anyone can shed any light.  

https://app.airdata.com/flight/bfcadba4ce4b176b73f20a66ce69589b

I did not touch the descent control. The drone should have appeared directly me at the same height that I left the final inbound turn from.

*edit*  I did not touch the descent control, only the forward control. The drone should have appeared directly over me at the same height that I left the final inbound turn from. Nothing explains the descent and nothing explains the impact with 26ft elevation still showing.

I estimate that the parts for the repair will be around $1500

Rook, where were you while all the 48 warnings were going off? With all of that empty space out there you could have just set her down and walked to it. Safety First! Ken.

Highasakite Posted at 1-31 02:43
Rook, where were you while all the 48 warnings were going off? With all of that empty space out there you could have just set her down and walked to it. Safety First! Ken.

Those warnings went off when it impacted the ground..

This is a classic case of CFIT. I don't know your background or knowledge of how these systems work but relying on the barometer at that altitude and speed is purely pilot error. The supposed accuracy of the sensor on the Inspire 1 is 10cm but no one with knowledge of these systems will actually believe that when manned aviation barometers aren't even that accurate. Your comments regarding the GPS readings are null, the aircraft doesn't determine altitude based on GPS readings, at least not in this circumstance.

Don't fly at such high speeds at altitudes that could lead to a CFIT mishap, it's basic risk management. Learn from this.

HyperSpectral Posted at 1-31 05:05
This is a classic case of CFIT. I don't know your background or knowledge of how these systems work but relying on the barometer at that altitude and speed is purely pilot error. The supposed accuracy of the sensor on the Inspire 1 is 10cm but no one with knowledge of these systems will actually believe that when manned aviation barometers aren't even that accurate. Your comments regarding the GPS readings are null, the aircraft doesn't determine altitude based on GPS readings, at least not in this circumstance.

Don't fly at such high speeds at altitudes that could lead to a CFIT mishap, it's basic risk management. Learn from this.

If it cannot maintain an altitude of +- 40ft, how is anyone supposed to use autopilot software? I would expect errors of no greater than +-5ft with a system like this... as it has demonstrated perfectly for the last 20 hours.

There is nothing to "learn from this" if we don't know what the cause of the baro channel to be off by 40ft is. Also, why was the aircraft content to report a height of 30ft just before impact without attempting to climb back to 40ft (or whatever the height was at the start of the home leg)?

Yeah fine, call if CFIT if it makes you feel superior since my finger was on the control and I was, therefore, controlling it. It still should have been flying level.

Rook Posted at 1-31 10:17
If it cannot maintain an altitude of +- 40ft, how is anyone supposed to use autopilot software? I would expect errors of no greater than +-5ft with a system like this... as it has demonstrated perfectly for the last 20 hours.

There is nothing to "learn from this" if we don't know what the cause of the baro channel to be off by 40ft is. Also, why was the aircraft content to report a height of 30ft just before impact without attempting to climb back to 40ft (or whatever the height was at the start of the home leg)?

I could give 0 fucks about what the causal factor was. In fact, I consider myself biased against DJI due to false claims to sell product and horrible standards for long term data on reliability of their systems. I get that this is a face threat to you, but you can't let that influence your assessment after the fact and ultimately your attitude moving forward.

Saying it can't maintain +-40 ft is exaggerated, as is saying it can maintain +-10cm.

CFIT can occur in manual flight or any programmed mode, the fact that you had your hands on the controls isn't the defining factor.

You can learn a lot from this, but if you can't even grasp the fact that you put the sUAS in a risk prone situation without proper mitigations then yeah I guess you can't learn anything from this.

I did learn from it though, your experience was a great reminder of the complications that arise from trusting your instrumentation. I was also reminded of the importance of risk mitigation when operating at high speed and low altitude, even if that risk mitigation is flying the sUAS in a sparse area in case of failure so only the sUAS is damaged. But we can still do better, why not brief yourself to maintain a higher level of situational awareness when operating in those circumstances. Were you 'ahead of the aircraft' when you brought it back at 40ft and 40mph? I know I wouldn't take my eyes off the screen or sUAS, especially since I know the limitations of these systems - everyone should.

Even beyond that, a fundamental understanding of how altitude is determined barometrically or via GPS would tell you that there could be significant difference in surface altitude at one point of the flat vs another. We don't have LiDAR surface models of the earth in <1cm resolution yet, let alone .5m, aside from the military anyway. 40ft of change in surface elevation isn't much and certainly isn't a large margin when you operate in an industry attempting to push for safety culture to be #1.  

Rook Posted at 1-31 10:17
If it cannot maintain an altitude of +- 40ft, how is anyone supposed to use autopilot software? I would expect errors of no greater than +-5ft with a system like this... as it has demonstrated perfectly for the last 20 hours.

There is nothing to "learn from this" if we don't know what the cause of the baro channel to be off by 40ft is. Also, why was the aircraft content to report a height of 30ft just before impact without attempting to climb back to 40ft (or whatever the height was at the start of the home leg)?

I have tested my Inspire starting at 30ft showing.......straight line and increase throttle to full.....and guess what!!?

She starts loosing altitude due to the angle of attack......IF you don't add some throttle....this happens everytime.   Flying an autonomous app, your bird should not ever see full throttle, so that's a moot point.

If you need to be that low at WFO throttle,  WATCH THE DAMNED THING!!!

Randy

OK, we are getting somewhere, if we keep going we may get to that point where we learn something interesting about the system because I read loads of threads about, "it fell out of the sky" and nobody ever comes up with a good reason as to why (other than "you are an idiot") and therefore zero learning takes place.

So let's talk about the height logic... If you are telling me that this thing isn't using GPS to determine height, then it must be using either an INS derived solution or a barometric solution. If it is the latter, then it probably has a static port somewhere which can become blocked or maybe confused at high speed. Is this part of a pre-flight check? I have never heard of it.  The indicated height was around 30ft when it impacted.

My point of "it cannot maintain 40ft" is based simply on the fact that it was at 40ft, I hit forwards, then it wasn't at 40ft. It impacted the dirt (and I was about 450ft away listening to it hit the dirt, to answer the question above about where was I when 48 warnings were going off).  At the impact point it should have read zero because...

To address the issue of Lidar scans of the ground to get accurate hight calls - valid point but in this case, million year old lake bed. Flatter than your average bowling lane. No measurements required. When it floods you get a beautiful 1" deep lake across the entire bed. It is actually pretty amazing - I have some footage of me flying over it a couple of years ago at about 3ft... and maintaining perfect height without touching the left stick.  Hmm, go figure.

@rwynant V1 - your test flight is actually very interesting and useful. What that tells me is that the flight software now goes into an apparent speed preference mode whereby it has tilted forward (looks like about -22 degrees from the .txt file) and tried to fly forwards at maximum speed by sacrificing height. Unfortunately, through what I would consider ... software, it hasn't maintained enough of a vertical component of lift to maintain height, well not after about the first 500ft anyway.  It started off level at full power, then randomly got to a point halfway home where it descended. What changed?

Bearing in mind these things are designed to be flown by non-professional pilot types it should probably be designed to stay level unless directed not to be. On all previous software loads, this was entirely capable of maintaining height at full throttle. If adding additional "up command" (I guess DJI call it throttle, but all controls change the motor speeds) enables it to maintain the aircraft level then the flight control software probably should have done that for me. That's how I would have designed it anyway. Kind of like my Mavic, that has zero issues.

The next test I would do, if I still had a working aircraft would be to try and, say, fly at 80% fwd throttle and see if it is capable of automatically maintaining height. Then the LEARNING POINT we can all take away is that if you fly at full throttle you cannot expect the aircraft to be able to maintain height due to too much lift being directed in the horizontal plane. Perhaps this one nugget of info will stop someone else repeating the same error and crashing their drone.  Yaaaay, we all win.

Rook, I am so sorry for the crash of your Inspire 1, I understand your frustration. Yet, if the drone is not within the warranty period, data analysis is not provided, and the customer may need to pay the repair according to the aftersales policy.
May I know if you had started a case? If so, you can provide me with your case number, I will double confirm it.

Rook Posted at 1-31 18:26
OK, we are getting somewhere, if we keep going we may get to that point where we learn something interesting about the system because I read loads of threads about, "it fell out of the sky" and nobody ever comes up with a good reason as to why (other than "you are an idiot") and therefore zero learning takes place.

So let's talk about the height logic... If you are telling me that this thing isn't using GPS to determine height, then it must be using either an INS derived solution or a barometric solution. If it is the latter, then it probably has a static port somewhere which can become blocked or maybe confused at high speed. Is this part of a pre-flight check? I have never heard of it.  The indicated height was around 30ft when it impacted.

Why don't you actually take the bits of information I provided and learn from them instead of taking it purely as a face threat? No one has directly or indirectly called you an idiot.

The curvature of the earth isn't really a factor at 8 inches per mile of elevation change but the so called flat lake bed probably isn't flatter than 8 inches per mile, but you know what? It's irrelevant.

I'd love to see DJI take responsibility for their systems like other aviation manufacturers and look into what caused your Inspire 1 to have trouble detecting actual altitude but you're still making excuses for the risk you took on that resulted in CFIT instead.  You don't even know what mode the system was in do you? Was the downward facing sensor even active? I'd bet not. You were likely in P-ATTI or F mode. Being aware of every sensor and flight controller mode and their limitations is the responsibility of the RPIC. You think we blame the manufacturer when pilots land with the gear up while the gear up annunciation is going off? No, yet it happens more often than you'd think.

I checked today......in P-mode,  at 40ft not touching the throttle stick, FULL FORWARD I lost about 3ft....but that's all in 1200ft of test.  I1 V1 new battery, if that makes any difference.

I did not test this in Atti-Mode....although I should have.....sorry.

rwynant V1 Posted at 2-1 15:09
I checked today......in P-mode,  at 40ft not touching the throttle stick, FULL FORWARD I lost about 3ft....but that's all in 1200ft of test.  I1 V1 new battery, if that makes any difference.

I did not test this in Atti-Mode....although I should have.....sorry.

@rwynant V1 - Many thanks for the test - that is exactly what I would have expected.

I never operate in atti-mode. Only P or F.  I also never turn off the downward position sensor though I am not entirely sure it would have done anything for me at this point.

DJI Thor Posted at 1-31 19:53
Rook, I am so sorry for the crash of your Inspire 1, I understand your frustration. Yet, if the drone is not within the warranty period, data analysis is not provided, and the customer may need to pay the repair according to the aftersales policy.
May I know if you had started a case? If so, you can provide me with your case number, I will double confirm it.[view_image]

Thor, thanks for the response.

I have started a case and I will DM you the case number. I have so far also given the same brush-off about warranty period and I would be totally cool with that IF you guys didn't constantly mess with the software in these systems so that their flight behaviors change. I don't have a choice if I update or not because, as I have found in the past, the thing will simply not even start the motors if the aircraft, controller, DJI Go app and iPad IOS are not all current.

Personally, I think that if DJI issues new software then it should take responsibility and re-hack the warranty period for software-specific incidents. Fine, if the hardware had failed and the prop fell off then DJI could rightly call "worn out, sorry", but not when it is as a direct result of new software.

If anyone smart wants to do their own assessment of the aircraft .txt file please DM me and I will send it to you

Rook Posted at 2-4 09:20
Thor, thanks for the response.

I have started a case and I will DM you the case number. I have so far also given the same brush-off about warranty period and I would be totally cool with that IF you guys didn't constantly mess with the software in these systems so that their flight behaviors change. I don't have a choice if I update or not because, as I have found in the past, the thing will simply not even start the motors if the aircraft, controller, DJI Go app and iPad IOS are not all current.

I am terribly sorry for the late response. I had replied the DM, please kindly check, thank you.