Does RTH work with Compass Error?

[chris pecchia]

If the compass error means you have no compass, the aircraft can't come home because it doesn't know where home is. Even if it knows its own GPS location and even if it knows the home point location, it still can't find home. It's like - what if you woke up and found yourself in a field in complete darkness. You know your house is exactly 100 meters due north, but you have no idea where north is. So you stumble off in a random direction hoping to get lucky...

For the first time today I had the same problem

Do you know if the RTH function is still working if the Phantom reports a Compass Error failure?

From time to time I get this error (maybe once every 20th flight) and I usually have to pilot the Phantom back using Atti mode and the live video feed. But what if I would lose the video link as well (e.g. because of the distance) or completely lose my orientation? Would it work to initiate RTH and get the Phantom to return to the home point - even with an active compass error? Or is "Compass Error + Loss of FPV" a certain killer combo?

//Tom

I had to same problem today first time everwe will fight seems like a bunch of people had the same problem today mine did the same thing I lost satellite signal but I was able to fly back home if I can't return to home I don't know if it would've worked very very scary

 

[ianwood]

The aircraft can see (much better than you.) It has GPS. It knows it's exact position all the time. It is not flying blind. Is that so difficult to understand?

Not knowing which way you're pointing is as good as blind. Imagine you move and you don't know if you wound up where you did because of your feet or the wind. I can simply assume it was all feet and determine my heading from it. But when I try again, I am going to miss again because there was almost definitely some wind in there. I can try to track that heading and determine a correction using the IMU but it's going to be good for only a few seconds at best before its garbage data. Two variable forces being applied neither of which are consistent.

So in anything but a dead calm or following wind, you'll continuously be missing the target. In a perfectly steady wind (which really doesn't exist outside of wind tunnels), you will fly off on a giant arc semi related to how much wind force is being exerted. It may get home after several hours of circling if it doesn't hit something first (which it will).

In a variable wind, you have to keep refactoring across two unknown forces that both leave you without knowing which way is which. So it's not really just a matter of attempting, missing and re-aiming. You have to keep re-guessing the wind corrected, IMU drifted heading every couple of seconds in order to figure it out. How does one do that every couple of seconds without going further and further off course?

Finally, I would never want my Phantom to drunkenly wander all over the place trying to find it's way home. The first sign of a compass error, I put it in ATTI and fly it home. It's really not hard. So this idea of a compass-less RTH for the Phantom is both not possible and pointless.

 

[syotr]

Have you ever flown a multirotor without the use of GPS, compass and a screen to show where it is at? Did it fly away? Were you able to bring it home?

 

[Shammyh]

Not knowing which way you're pointing is as good as blind. Imagine you move and you don't know if you wound up where you did because of your feet or the wind. I can simply assume it was all feet and determine my heading from it. But when I try again, I am going to miss again because there was almost definitely some wind in there. I can try to track that heading and determine a correction using the IMU but it's going to be good for only a few seconds at best before its garbage data. Two variable forces being applied neither of which are consistent.

So in anything but a dead calm or following wind, you'll continuously be missing the target. In a perfectly steady wind (which really doesn't exist outside of wind tunnels), you will fly off on a giant arc semi related to how much wind force is being exerted. It may get home after several hours of circling if it doesn't hit something first (which it will).

In a variable wind, you have to keep refactoring across two unknown forces that both leave you without knowing which way is which. So it's not really just a matter of attempting, missing and re-aiming. You have to keep re-guessing the wind corrected, IMU drifted heading every couple of seconds in order to figure it out. How does one do that every couple of seconds without going further and further off course?

Finally, I would never want my Phantom to drunkenly wander all over the place trying to find it's way home. The first sign of a compass error, I put it in ATTI and fly it home. It's really not hard. So this idea of a compass-less RTH for the Phantom is both not possible and pointless.

Wowza! This thread is still going? Did people not bother to read the previous, rather comprehensive, discussion?

As Ianwood has pointed out several times now, the P3 compass and the current algorithm suite are insufficient to allow for RTH in the event of compass failure. Period. End of story. If you lose compass, you're flying Atti. Prepare accordingly.

As I pointed out, along with others, there may be other technical solutions to allow for RTH without a compass, but those discussions are probably better held on Rcgroups or other places where people are using open source flight controllers and DIY hardware (plus, those forums skew more towards engineers, whereas here, we skew more towards end-users). It's extraordinarily unlikely DJI will implement a drastically improved RTH alg for the P3. That's the deal with closed source hardware. That said, I still adore my P3!

I might disagree with Ianwood, in that I think it's technically possible, but very much agree that we will never see such a feature in our birds.

Also, as a side note, almost every car in the past several years that has integrated gps also has a compass. Whether or not your car gives you access to its data independently is irrelevant. To test it, just look at your GPS screen and try turning slowly in a circle. You'll see immediately if it tracks you turn.

 

[MikeyOnline]

So in anything but a dead calm or following wind, you'll continuously be missing the target. In a perfectly steady wind (which really doesn't exist outside of wind tunnels), you will fly off on a giant arc semi related to how much wind force is being exerted. It may get home after several hours of circling if it doesn't hit something first (which it will).

...

Finally, I would never want my Phantom to drunkenly wander all over the place trying to find it's way home. The first sign of a compass error, I put it in ATTI and fly it home. It's really not hard. So this idea of a compass-less RTH for the Phantom is both not possible and pointless.

But how is that possible? How can you possibly fly it home in ATTI mode with all the wind and other variables without taking a "drunken arc" and flying for hours? You said "Not knowing which way you're pointing is as good as blind"... You mean to tell me that you can tell which way you are pointing from that dot 300 meters away by looking at the dot in the sky without any RC input? Of course not. You give it a little forward pitch on the right stick and watch where it goes. Given what you see, you make adjustments until you determine how to bring it home: same exact thing this simple non-compass RTH algorithm can do. And the AC has a big advantage you don't have in that it actually knows where the AC is and which direction it is headed (after pitch is applied and movement starts). Flying in ATTI mode, you have to get its direction less directly and you will have to perform more maneuvers to do it if you are using LOS and not telemetry.

Wind, IMU, gyro accuracy, none of that matters! This is really a very simple problem. It's the same problem you face when you have to fly home in ATTI mode: both you and the AC have the same information to work with (where the AC is and which direction it moves when you give it some forward pitch). There are multiple ways to do it but I find this the simplest (for flying the standard/default mode 2):

• +25% pitch on right stick and hold
• How did the P3 move, either by watching it in the sky or on telemetry?
• Only 3 possibilities here if you are flying LOS watching the dot in the sky that is your P3:
• Dot isn't moving: you're either flying toward or away. To find out which, hold a little yaw (left or right) on the left stick until...
• Dot is moving left: hold slight left on the left/yaw until the dot is no longer moving left
• Dot is moving right: hold slight right on the left/yaw until the dot is no longer moving right
• Keep the pitch forward while repeating the previous two steps, keeping the AC from moving left/right relative to your LOS as it approaches home

Once you've performed that procedure, your P3 is headed toward home. It makes no difference how accurate the IMU is or which way the wind is blowing. If you have a strong wind coming from one side, the P3 will be "dogging" it's way home, slightly sideways. Doesn't matter. Everything is relative because you simply set the yaw to get your line home: it doesn't mean the P3 is pointing home, only that it is headed home: your pitch+yaw is what is required to make it run that line. If, on the way home, you see it starting to drift to your left or right a little, make fine adjustments with the left stick to stop the left/right drift as you continue to pitch forward on the right stick. Your track home flying manual ATTI should look like an "S" around the straight home line: basically follows the home line but with slight variations side to side along that line.

A computer algorithm doing the same thing is guaranteed to follow that straight home line closer than you did manually because it can see where the AC is accurately, which way it is moving, and the moment it veers even slightly off the home line, it can apply the slight adjustments needed to keep it close to that line. And again, wind, IMU or gyro accuracy, etc. don't matter. All it has to do is pitch forward, adjust yaw until the AC is headed toward home based on its GPS movement, and then make "micro adjustments" to the yaw from that point whenever it veers to one side or the other of the home line along the way. In a couple minutes, your P3 will be overhead and you can manually land.

Shammyh has a good point: why bother? I agree: if I can see it, I personally would probably just fly it home even if we had this algorithm. But there are a lot of situations (particularly if your compass has failed) where you may lose line-of-sight. And we all know from watching the videos posted on this forum that with LightBridge, we don't all follow the FAA guidelines to keep LOS all the time. In such cases, a simple algorithm like this could mean the difference between loss of the AC and getting back close enough to home to (re)take control.

Mike

 

[Mark The Droner]

...... You mean to tell me that you can tell which way you are pointing from that dot 300 meters away by looking at the dot in the sky without any RC input? Of course not. You give it a little forward pitch on the right stick and watch where it goes. Given what you see, you make adjustments until you determine how to bring it home: same exact thing this simple non-compass RTH algorithm can do. ...

Sorry but it's not the same exact thing - it's not even close. As ianwood has repeatedly tried to point out, if your algorithm attempts to push the craft forward five feet, and the gusty wind blows the aircraft back 10 feet, how is the algorithm going to know what is going on?

 

[MikeyOnline]

Sorry but it's not the same exact thing - it's not even close. As ianwood has repeatedly tried to point out, if your algorithm attempts to push the craft forward five feet, and the gusty wind blows the aircraft back 10 feet, how is the algorithm going to know what is going on?

And as I (and others) have repeatedly pointed out, you know what's going on by watching where the AC goes by it's GPS location. What would you do if you tried to move it in one direction and the wind pushed it in the opposite? You might make an initial mistake and assume you were facing the wrong way but a few changes to yaw and you'll see what is going on. The algorithm is no different and unless GPS goes out too, it has the same information that you have on the ground flying in LOS ATTI mode, only it's computations will be better than yours. All it really needs to do is go up on the right stick to pitch forward: that gives it a single thrust vector. You change the angle of that vector with yaw and after (at most) 10 seconds, you've adjusted your thrust level and angle at the correct angle to start heading home. If you need to fly due north and there is a strong wind from the west, your thrust angle will be northwest but the AC will be flying due north, and so on. If wind gusts are so variable and unpredictable that you can't figure out which way to point your thrust vector to head toward home, you're not flying it home manual ATTI either!

In your example, you might be facing north (you don't know this yet) and you apply 25% pitch forward. If the wind was blowing steady from the north, you would have already been moving south prior to the application of pitch and +25% pitch would slow your southern movement. But let's take worst case scenario and assume that some roque gust of north wind hits you just as you apply +25% pitch facing north and you begin to move south. The AC might make an initial mistake and think it's facing south (just as you would if you were flying ATTI), but after applying a steady yaw for a full rotation of the AC (again, at most, 10 seconds), you'll realize the error and get your bearings. And again, nothing can change the fact that both you and the AC have the same control and same information available about location and direction. Neither one can know orientation without a few stick movements to deduce that while watching movement, but the process is the same. And it's not a very difficult process at that.

Mike

 

[syotr]

Sorry but it's not the same exact thing - it's not even close. As ianwood has repeatedly tried to point out, if your algorithm attempts to push the craft forward five feet, and the gusty wind blows the aircraft back 10 feet, how is the algorithm going to know what is going on?

The aircraft can easily know wind speed and direction. Just let it drift for a second. The GPS knows starting and ending points so it is an easy calculation. Combine this vector with the forward flight one and you have your heading. It is simple math.

 

[Mark The Droner]

And as I (and others) have repeatedly pointed out, you know what's going on by watching where the AC goes by it's GPS location. What would you do if you tried to move it in one direction and the wind pushed it in the opposite? ...

Sorry but you missed a key word: "gusty". I even underlined it.

So you make a few changes in yaw and you still don't know what's going on because wind is not constant. That's the problem.

Could it be you guys don't know what gusty means?

 

[MikeyOnline]

Sorry but you missed a key word: "gusty". I even underlined it.

So you make a few changes in yaw and you still don't know what's going on because wind is not constant. That's the problem.

Could it be you guys don't know what gusty means?

No, I even gave the "worst case scenario" with a gust approaching just as you make a change, making it look like your change (to thrust or angle) did something it didn't really do. Over time, that all comes out in the wash, just as it does when you are flying it home yourself. If you can watch the AC and get it headed in your general direction with a few stick movements and then compensate for gusts that may push it off course as you come home, the AC can do it even more accurately than you can. The only way it wouldn't work is if the wind or gusts are higher than the thrust capability of the AC... in which case, it's a goner anyway. This isn't about deducing a thrust angle with one movement and then flying blindly. It's a continuous loop that, over time, is able to keep making adjustments until the GPS coordinates are moving toward home. Same thing you do when you are flying ATTI in gusts.

And as syotr pointed out, if you wanted to discover wind speed and direction, just level the AC via the IMU and hover with it level for a few seconds. Watch how it moves on GPS and that's your wind speed and direction. Doesn't matter that it's an average or that there are gusts. Those can be compensated for during flight, just as you would if you were flying it: course corrections.

Mike

 

[syotr]

How do you drive your car down the road? When it drifts, you steer a little bit in the opposite direction. You make constant corrections to stay on track. You don't worry about your exact heading or turns in the road. You correct as you go.

 

[johnnyu]

Geocachers use GPS only and find pinpoint locations give or take a few feet on GPS alone. A compass then would only be useful to indicate which way your pointing if you were standing still. Take two GPS vectors and that would tell you direction of travel and coordinates based on all the GPS data coming in..

 

[dgd3]

Is the orientation triangle arrow on the DJI Vision app a direct result from the Phantom compass?

 

[BudWalker]

Is the orientation triangle arrow on the DJI Vision app a direct result from the Phantom compass?

Not exactly. The orientation triangle comes from the flight controller's Yaw estimate. The Yaw estimate is derived from the compass magnetometers and then corrected for pitch and roll. Almost always, the Yaw estimate is the direction the A/C is pointing. But, in unusual circumstances the Yaw estimate can be incorrect.

 

[dguth66]

The DJI tech guy I spoke to some days ago, was clear that none of the smart navigation functions will work reliably if there is a compass error. Compass errors are extremely serious malfunctions and should not be taken lightly. You should immediately switch to Atti mode and enjoy the bumpy ride back home, flying by the seat of your pants (and the FPV image on the screen). Once you are near your home point and you can actually hear and see the drone, you can try switching to P mode again and, if the screen turns green, land normally. Yay!

However, if you initiate RTH in a compass error situation or if the drone loses the RC signal and self-engages the failsafe RTH, it will not try to use any kind of hot/cold algorithm - it will simply assume that it knows its correct position and heading, and will fly accordingly. You may - or you may not - end up where you intended. If this is poor programming or not, I cannot tell.

For me, this explanation will do.

//Tom

Great advise ... and reminder to what u said before ... be careful and mindful when doing a compass cal (read the manual, understand it, read it again and realize there is no need to do it before every flight ... only if flying in a diff area many miles away from last cal pos).

I lost a perfectly good (new P4) because of some internal nav error where RTH did not work either when commanded nor did it work when signal lost - in auto ... by work I mean "stop flying away" and change it's course to come back. Because I did not have the stick experience & happened so fast at the time I did not think to switch to ATTI mode ... if I did perhaps I'd have that new P4 in the now empty case.

Ole saying ... fail to plan - you plan to fail !

 

[N017RW]

The advice to limit compass cals is a good one.

However there are no set requirements for doing so other than observed flight characteristics or prompting from the App..

I have traveled 600 mile from my last cal location and had no issues at that location or upon return.
(My last cal on my P2 was about 1.5 years ago, last flight about 2 weeks ago.)

 

[curiosity]

The aircraft can see (much better than you.) It has GPS. It knows it's exact position all the time. It is not flying blind. Is that so difficult to understand?

No offense, but at this point, I guess it's time to understand the fact that it might not be as easy as it is to us for some other people to understand.

We can't expect everyone to see a solution like another one can do. This has always been the case with inventions and their inventors - some were able to "invent" while others thought it was not possible; it's historically proven .

 

[ianwood]

We can't expect everyone to see a solution like another one can do. This has always been the case with inventions and their inventors - some were able to "invent" while others thought it was not possible; it's historically proven .

Yeah right. More like some of us have actually led R&D teams to develop fused inertial systems in navigation applications and know the limitations of consumer grade IMUs all too well. I keep saying it like a broken record:

• Tell me how to correct for winds that change velocity every second and can vary direction by 45 degrees within 50ft.
• Tell me how to hold a heading for more than 15 seconds using only a sensor that without any reference adjustment will produce garbage values in half that time.
• And we haven't even discussed the accuracy of a GPS heading being directly proportional to the distance covered.

Three really good reasons why this is a bad idea.

But the best reason why this should never exist is the simple ability to fly home under pilot control using ATTI.

 

[dguth66]

How ??

The Northrop Grumman X-47B ...also known as the "Salty Dog"
Sorry could not resist

 

[syotr]

You know some folks said we couldn't fly to the moon. Some swear it was faked rather than admit they were wrong.