Copter pauses/stops during Auto flight

I’m using a DJI S900 copter with Pixhawk. I’ve noticed the copter “pausing”, or at least wide swings in speed from the planned speed. This seems coupled with the inability to hold the planned altitude very well. Anyone else seen this, or know why this might be occurring? I’m attaching a sample log. Vibration levels seem reasonable. This flight breached the altitude fence (10m higher than mission alt) and also tipped over on landing doing a good bit of damage.

2017-01-27 14-06-21.bin (4.0 MB)

I don’t have the software in this computer to be able to look at your bin file right now. But a couple things - I think in copter the aircraft will “twitch” or "pause slightly at each way point. Unless you use spline waypoint, then it doesn’t do that. This, of course, will cause variation in speed.

The other thing is that if you have altitude change between waypoints, and the aircraft cannot make the target altitude at the desired ground speed within the vertical acceleration limits you have set, the software will adjust the speed accordingly.

Finally, winds aloft is an issue. If your 'copter does not have enough power to maintain ground speed with changing winds aloft, within the frame tilt limits you have set, speed will vary.

This is just a personal preference, but I never trust auto-landing, especially in unknown terrain, and always take manual control at the end of a mission and land in Stabilize flight mode so I have control of the machine if it lands on uneven ground and may tip over or experience a prop strike. I think the S900 has kind of a high CG and is probably prone to tipping over even if the landing gear gets caught on some grass while the aircraft is fighting ground effect. So I prefer full manual control on both takeoff and landing for missions.

These are just a few ideas off the top of my head. When I get home later I can then take a look at your bin file on one of my laptops that has APM Planner2 in them.

I see a couple things in the bin file:
1.) The actual frame tilt appears to be limited to ~15 degrees. The autopilot wanted up to ~45 degrees, and thruout the whole flight actual pitch varied considerably (less) than desired pitch. This is obviously going to cause a problem with ground speed because a 'copter has to be able to get pitch in order to have forward (or any direction) flight.

I would look at your max allowed frame tilt parameter to see what it is set to.

2.) The RTL breached the altitude fence. Likely your RTL altitude is set higher than the fence allowed. So the FC err’d on the side of safety and climbed to the RTL altitude instead of obeying the fence. But the aircraft wasn’t moving. So it descended back below the fence altitude and then proceeded to the crash site.

The Pixhawk did detect the tipover at the end of the flight. Sounds like you got damage before it shut the motors down. That is the risk you take with auto-landings. Sometimes they go perfect. Sometimes ground effect or a wind gust will cause it to correct, the landing gear snags, and if the aircraft is top heavy and has long legs, over it goes. IMO the only reason to use an auto-landing is for emergency landings when something bad happens and during the landing the aircraft descends below VLOS so you can’t land it manually.

It would be interesting to play back the tlog from the flight if you have it available, to verify my suspicions on pitch attitude vs speed and altitude changes.

Chris–thank you for the analysis. I checked, and ANGLE_MAX is 4500 (45 degrees).

There’s a related post in copter 3.4 (Quad stop in auto mode), where I also appended some telemetry log graphs. The thinking seems to be that it’s altitude control related, and perhaps vibrations are the root cause. Still studying logs and posts found in that thread…

I’ll take a look at that thread too. It seems to me I had this problem with one of my Tri’s some time back, and I don’t remember what I did to fix it. When ever I change my params I always save the old param file and date it with a comment as to what I did. I may have to search back thru those to see if it was the same issue.

I guess I don’t really see anything out of whack as far as X, Y and Z vibration levels. Short of where the props hit the ground, sort of raising havoc with things.

It would be nice to have the tlog file from your ground station so I could “fly” the flight by playing the tlog back and get a feel for what the aircraft was doing. The data flash logs are nice for some things. But the telemetry data often contains more easily visualized info about the flight, loads on the motors, heading vs pitch, roll and wind, etc… I have my doubts it was a vibration issue. That looks pretty tame from this end. It confuses me as to why the aircraft came to almost a dead stop between commands 7 & 8. The tlog would contain the flight plan as downloaded from the aircraft, and what it was supposed to do, vs what it actually did do.

This appears to be solved. See link above to “Quad stops in auto mode”. It appears to have been caused by high vibration levels. I used a vibration isolation mount and Zeal 3mm gel pads, both available on Amazon.