Failsafe RTH

[ChopperLinc]

I recently posted about a drone crash where it seems the aircraft suddenly fails mid flight and fell out of the sky. While I have a warranty claim in, I’m still trying to figure out if I did something wrong.

Because I have no drone at the moment, reading the manual again is the next most exciting thing.

The section on Failsafe RTH states the the aircraft uses the Forward Vision System to create a map of its path, and if the compass is working and the home point is recorded, RTH will engage after two seconds of lost signal.

If the vision system doesn’t work at night, does that mean Failsafe RTH won’t work at night because it can’t create the real-time flight path? Also, is GPS alone not sufficient to bring the drone home?

Seems odd that given the amount of focus on mag interference that the Failsafe would be based on something as unreliable as the compass, and I don’t understand the need for a real-time fight path when all the lat and long data is recorded and available to the aircraft.

Any insights?

 

[Meta4]

The section on Failsafe RTH states the the aircraft uses the Forward Vision System to create a map of its path, and if the compass is working and the home point is recorded, RTH will engage after two seconds of lost signal.

If the vision system doesn’t work at night, does that mean Failsafe RTH won’t work at night because it can’t create the real-time flight path? Also, is GPS alone not sufficient to bring the drone home?

DJI is a little sloppy with their communications.
It doesn't mean that Failsafe RTH won't work at night at all.
It just means that it won't be able to retrace to the last point where it had good signal. and RTh will work the good old fashioned way of climbing to RTH height and coming home.

Seems odd that given the amount of focus on mag interference that the Failsafe would be based on something as unreliable as the compass, and I don’t understand the need for a real-time fight path when all the lat and long data is recorded and available to the aircraft.

It doesn't seem odd at all.
The compass is not unreliable at all. But some flyers can't be relied on to avoid launching from steel or reinforced concrete surfaces and messing up their compass.
It's not just RTH that's based on compass.
Your drone can't fly straight at all without the compass.
Lat and long data is no use without having a compass to say which direction to get from point A to Point B.

 

[sar104]

I'm also not sure why the vision system is necessary to retrace the aircraft's track - the lat/long track plus altitude is its track record and doesn't rely on any vision data as far as I know. The forward vision system is not going to be recording much useful data in the absence of nearby obstacles.

I wonder if the retrace really doesn't work at night or if that is another DJI miscommunication - like the compass calibration advice.

 

[ChopperLinc]

I'm also not sure why the vision system is necessary to retrace the aircraft's track - the lat/long track plus altitude is its track record and doesn't rely on any vision data as far as I know. The forward vision system is not going to be recording much useful data in the absence of nearby obstacles.

I wonder if the retrace really doesn't work at night or if that is another DJI miscommunication - like the compass calibration advice.

What is the calibration advice that’s incorrect? Seems like the role of the compass is misunderstood based on all the differing opinions here.

 

[ChopperLinc]

I'm also not sure why the vision system is necessary to retrace the aircraft's track - the lat/long track plus altitude is its track record and doesn't rely on any vision data as far as I know. The forward vision system is not going to be recording much useful data in the absence of nearby obstacles.

I wonder if the retrace really doesn't work at night or if that is another DJI miscommunication - like the compass calibration advice.

Heading can’t be determined by movement between two coordinates?

DJI is a little sloppy with their communications.
It doesn't mean that Failsafe RTH won't work at night at all.
It just means that it won't be able to retrace to the last point where it had good signal. and RTh will work the good old fashioned way of climbing to RTH height and coming home.

It doesn't seem odd at all.
The compass is not unreliable at all. But some flyers can't be relied on to avoid launching from steel or reinforced concrete surfaces and messing up their compass.
It's not just RTH that's based on compass.
Your drone can't fly straight at all without the compass.
Lat and long data is no use without having a compass to say which direction to get from point A to Point B.

Still doesn’t make sense to me. Seems like a fairly simple calc. Record first point, forward input, record second point, difference gives heading. No?

 

[sar104]

What is the calibration advice that’s incorrect? Seems like the role of the compass is misunderstood based on all the differing opinions here.

The advice that calibration is necessary when changing locations.

I'm sure that most people broadly understand the role of the compass. The subtleties of the sensor fusion scheme, in which the compass is not the primary source of fast yaw data after the IMU is initialized, are not widely understood, but that's a different issue.

I'm not sure what you mean by "differing opinions" - this is a well-understood field of engineering and opinions don't really come into it. If you are up to understanding the physics and maths then a simple literature search reveals much of the research underlying these kinds of control systems, and while the exact implementation details are certainly proprietary, the concepts are not.

 

[sar104]

Heading can’t be determined by movement between two coordinates?


Still doesn’t make sense to me. Seems like a fairly simple calc. Record first point, forward input, record second point, difference gives heading. No?

Agreed - as I said, the recorded flight track should be sufficient to retrace the path. As an analogy, the FC can obviously fly a waypoint mission at night - it doesn't need the vision system to navigate.

 

[ChopperLinc]

I'm not sure what you mean by "differing opinions"

Maybe better language is differing levels of understanding, to your point. Meta seems to think that the compass it totally necessary for RTH, but I hear you saying that it should be able to rely on GPS.

While, admittedly, my understanding of the science is probably the most rudimentary here, I'm just trying to answer simple questions about the Mavic 2 implementation, like, "If I fly at night, will my FRTH fail," or, "If upon engagement of FRTH I experience magnetic interference, will my drone get lost."

As Meta said, some of the Mavic language is clunky at best.

 

[sar104]

Maybe better language is differing levels of understanding, to your point. Meta seems to think that the compass it totally necessary for RTH, but I hear you saying that it should be able to rely on GPS.

While, admittedly, my understanding of the science is probably the most rudimentary here, I'm just trying to answer simple questions about the Mavic 2 implementation, like, "If I fly at night, will my FRTH fail," or, "If upon engagement of FRTH I experience magnetic interference, will my drone get lost."

As Meta said, some of the Mavic language is clunky at best.

Oh - I did not mean to imply that the compass is not necessary for RTH. No navigation at all can happen without the compass even if the IMU yaw is correctly initialized at the start of the flight, since the rate gyro drift is too large and the inertially-computed yaw would rapidly become incorrect. Without compass data the FC switches to ATTI mode. To go back to your example, the required heading is easily computed from target coordinates but without the compass the FC doesn't know which direction it is facing, and so doesn't know in which direction to apply thrust to move towards the target coordinates. It could attempt a trial and error solution but, with the relatively slow data rate from the GPS unit, it would end up staggering around looking drunk. It doesn't even know how to combat drift due to wind in that situation - hence the switch to ATTI mode.

 

[ChopperLinc]

Got it. Thanks for the explanation.

 

[DanMan32]

Vision is used so that it is less likely to hit something on its way back. No different than regular flight with APAS on.

 

[sar104]

Vision is used so that it is less likely to hit something on its way back. No different than regular flight with APAS on.

So without vision (i.e. at night) it won't attempt that? It seems it would be more likely to encounter an obstacle heading directly home, since that is going to be the most probable reason for the disconnect in the first place.

 

[Kilrah]

That's not an issue if you've set your RTH height correctly.

 

[sar104]

That's not an issue if you've set your RTH height correctly.

If we could rely on the pilot to set everything up correctly then disconnects would almost never happen to begin with.

 

[ChopperLinc]

If we could rely on the pilot to set everything up correctly then disconnects would almost never happen to begin with.

So, if you’re further inclined to humor a newbie, what are the most important things to consider if trying to prevent disconnects?

 

[sar104]

So, if you’re further inclined to humor a newbie, what are the most important things to consider if trying to prevent disconnects?

Based on the events that get reported here - flying too far (obviously), especially in noisy RF environments, flying behind obstacles (buildings, hills etc.) using mobile devices that seem not to play nicely with the RC (mostly Android devices seem to cause that problem) and using low-quality USB cables to link the device to the RC. It's notable how many times the statement is made that the problem had occurred before but nothing had been changed to fix it.

 

[ChopperLinc]

Based on the events that get reported here - flying too far (obviously), especially in noisy RF environments, flying behind obstacles (buildings, hills etc.) using mobile devices that seem not to play nicely with the RC (mostly Android devices seem to cause that problem) and using low-quality USB cables to link the device to the RC. It's notable how many times the statement is made that the problem had occurred before but nothing had been changed to fix it.

I’ve heard mention that one shouldn’t use the Stock cable that comes with the controller. Is that your sentiment as well?

 

[sar104]

I’ve heard mention that one shouldn’t use the Stock cable that comes with the controller. Is that your sentiment as well?

I use the stock cable (iOS) and haven't had any problems. I have had disconnects with a CrystalSky but I was running beta software, so it's not clear to me whether the problem was the software or the cable.

 

[Meta4]

I'm just trying to answer simple questions about the Mavic 2 implementation, like, "If I fly at night, will my FRTH fail,"

RTH will work just fine in the dark (as long as your RTH height is appropriate to clear any obstacles that may be between the Mavic and homepoint.)

"If upon engagement of FRTH I experience magnetic interference, will my drone get lost."

Magnetic interference isn't something you should encounter up in the air.
At normal flying heights, your compass won't detect the magnetic fields associated with iron or steel objects on the ground.
You would have to be flying only a few feet away from a steel roof to see compass issues when flying.
Just avoid launching from magnetically dirty locations .. steel surfaces, reinforced concrete etc.

I’ve heard mention that one shouldn’t use the Stock cable that comes with the controller.

That's an example of the myths and misinformation are to be found in the forums.

 

[ChopperLinc]

Thanks, Sar and Meta for all the helpful information.