FORCE LANDING ?!

[msinger]

I'm assuming that downward sensor means OSD.sWaveHeight?

Yes.

I'm not so sure that there is a problem with the sensor, at least one that could be fixed. On the P3 the vpsHeight is quite noisy and gets a lot of spurious readings when the P3 is above 10 meters. Seems like I've read that the problem is worse over water, which was the case with this flight.

That all sounds reasonable. Perhaps DJI needs to tweak the code that is determining when to trigger the forced landing mode. And, until they make any changes (if they do), the OP could disable the downward sensors when flying over water.

 

[Ed209]

That all sounds reasonable. Perhaps DJI needs to tweak the code that is determining when to trigger the forced landing mode. And, until they make any changes (if they do), the OP could disable the downward sensors when flying over water.

I normally fly with VPS off on the P3 and have done the same with the Mavic at first. I found that the auto-landing with all of the nannies on is a bit unnerving since you don't really have control once it starts it's final descent. I think I will be turning all of them off unless I am flying in some really tight quarters.

 

[BudWalker]

I normally fly with VPS off on the P3 and have done the same with the Mavic at first. I found that the auto-landing with all of the nannies on is a bit unnerving since you don't really have control once it starts it's final descent. I think I will be turning all of them off unless I am flying in some really tight quarters.

You could also avoid full back on the throttle as it seems that full back for > 1 second (and OSD.sWaveHeight > 0.0) is required to trigger ForcedLanding. And, you can un-trigger ForcedLanding by full forward on the throttle.

It's just me but I don't want to disable the downward sensor. I've begun to trust the auto land feature. I always hand catch the P3 because I'm afraid of a tip over. But, the Mavic has a lower center of gravity and I'm not so concerned about a tip over. Hand catching has it's advantages but the props are awfully close to my hand with the Mavic.

 

[Ed209]

You could also avoid full back on the throttle as it seems that full back for > 1 second (and OSD.sWaveHeight > 0.0) is required to trigger ForcedLanding. And, you can un-trigger ForcedLanding by full forward on the throttle.

It's just me but I don't want to disable the downward sensor. I've begun to trust the auto land feature. I always hand catch the P3 because I'm afraid of a tip over. But, the Mavic has a lower center of gravity and I'm not so concerned about a tip over. Hand catching has it's advantages but the props are awfully close to my hand with the Mavic.

I think for now I'll keep it on and see if my typical flying style triggers it often. At least I know what to expect and how to handle it. I agree, the auto land does seem pretty solid. I think there is more confidence in it because the Mavic hardly drifts when hovering low to the ground. Even during windy days, I have put it down exactly where I wanted it. Maybe it's just that the uncomfortable thing is getting used to something new.

 

[Ed209]

Happened again just a few minutes ago. I suspect glare from the sun in the sensors to be the cause. I was flying from my house (not over water) this time and coming back to land. Was maybe 200ft up and the sun was hitting the camera/sensors from the side. Full left stick up aborts the landing.

 

[msinger]

Did you have the downward sensors turned on again? If so, did you check out your flight log to see if the sensors were detecting the ground when the Mavic was 200 feet up?

 

[Ed209]

Everything was on. I'll take a look at the logs now.

 

[msinger]

If you need help reading your log, please attach the TXT file here.

 

[Ed209]

If you need help reading your log, please attach the TXT file here.

I'm trying to do the work myself to learn a little. I'm looking at the time of the Forced Landing event and everything looks right except I don't see the flyCState change to reflect forced landing. In fact, I only see when it switches to ATTI for that split second when going into Sport mode. Am I missing something or are you and Bud working with an updated version? I'm using 0.9.9.

 

[msinger]

I don't use that tool. I normally just look at the CSV data. The "OSD.sWaveHeight" column in the CSV shows the height detected by the downward sensors.

 

[Ed209]

I don't use that tool. I normally just look at the CSV data.

Which column? Altitude(feet) says 98.8. Is that what you'd look at?

 

[msinger]

OSD.sWaveHeight

 

[Ed209]

OSD.sWaveHeight

Ahhh thanks, I was looking at the CSV downloaded from your site. Just exported a CSV from CsvView. OSD.sWaveHeight reads "0". It reads 0 except during takeoff and the actual landing. Is that normal?

 

[msinger]

I don't show all of the data in my CSV. It's still a work in progress.

That value should be 0 when the Mavic is too high for the downward sensors to detect the ground.

 

[Ed209]

That's interesting then. So then it's not getting a false reading from the glare and thinking it's near the ground I take it? Wonder what could be causing it. I tried to replicate it but I moved the camera gimbal and I didn't get it just right to replicate I think. And I confirmed that tapping the red "X" on the screen to cancel the landing does nothing. Just ascending full throttle cancelled it.

 

[BudWalker]

Ooops. I added the definition for OSD.sWaveHeight in my version without distributing it so others could use it. I didn't think anybody would actually want to do this. But, it's "field upgradable" Attached is a .zip containing the file txtSignals.xml use that to replace the txtSignals.xml located where CsvView is installed.Sorry about that.

 

[msinger]

That's interesting then. So then it's not getting a false reading from the glare and thinking it's near the ground I take it? Wonder what could be causing it. I tried to replicate it but I moved the camera gimbal and I didn't get it just right to replicate I think. And I confirmed that tapping the red "X" on the screen to cancel the landing does nothing. Just ascending full throttle cancelled it.

It would be nearly impossible to make sense of it without seeing the data myself.

 

[Ed209]

Ooops. I added the definition for OSD.sWaveHeight in my version without distributing it so others could use it. I didn't think anybody would actually want to do this. But, it's "field upgradable" Attached is a .zip containing the file txtSignals.xml use that to replace the txtSignals.xml located where CsvView is installed.Sorry about that.

Thanks Bud appreciate it

 

[BudWalker]

If you're feeling brave you can try this after you've installed that txtSignals.xml. Unzip the attached which will yield vps-fl.pml. Import that into CsvView by doing File->ImportPlayer. That should bring up the SigPlayer I've been using to look at your stuff.

 

[Ed209]

Thanks I will do that a little bit later tonight. Gotta run out for a bit.