Fixing DJI's Compass Problem

[Andy T]

After a few months of trying to initialize a response, It has now been three weeks since we first heard back from DJI and they sent out the beta. That is plenty of time to get the fix implemented and released to the public if they were working diligently on it. It is clear they are not.... at least not as fast as we should be demanding to accomplish our original goal!

I don't have a dog in this fight (I don't get the hook and have one of the first videos of a P1 showing straight flight in a high-dec location), but I did want to chime on the statement above. I work directly with multiple teams on public facing web and mobile services, software and operating system launches, and the above cannot be farther from the truth. Software release management is complicated process, and what seems like a simple change on the outside more often than not has multiple dependencies and cascading effects, all of which need to be carefully designed, documented, and tested against, including unit, integration and/or regression testing, verification and validation, etc. And this is just standard industry practice, not even taking into account a scenario where a failure could end up endangering or destroying property or causing physical injury. And this error is geographically-distributed and variable as well, meaning iterative changes require coordination of testing and validation across global regions.

I'm not saying it's wrong to push a manufacturer to prioritize a fix, but IME the statement above is inconsistent with basic release management methodologies for software or embedded systems.

We are talking communist China here where not that long ago the management of change procedure read something like this "if you don't get it right we are going to take you out back and shoot you"

I have witnessed even here where a company claims to be state of the art and their systems were laughable

 

[landonkk]

I understand your frustration but, just for the record, I don't regard 3.05 as a fix, and I'm not using it to fly except for beta testing. I only have it installed on one of my P2s for that reason. I believe that if they released it now then it would be very premature and that it would lead to a landslide of further complaints.

As an interim solution, you may find that rotating the compass by twice the local declination is a better fix, as several users have described.

By all reports Beta 3.05 at the very least improves the declination issue except for a few minor issues and even those are likely present in the latest public release. How would it cause a "landslide" of complaints?

I will say this again, as Lenny has echoed my thoughts, Project Drunken Sparrow's GOAL was to get a FIX to the magnetic declination error issue. We still do not have one. We are making progress because there has been a response, but why should we just shut up and be quiet before our goal is reached? So they are working on it.... Great. Work faster! The fact that DJI is leisurely dealing with this is in the first place is because we raised enough of a stink they felt they had to. Logic would indicate that if we kept the pressure on, they might work a little faster.

The psychology of perception is also at play here. Those that are included in the testing feel obligated to defend DJI because the perception is they are now part of the team working towards a solution and it is only natural to defend yourself. And those who are waiting for a fix now perceive the testers as part of the problem because they are working with DJI and won't release what we have so far to everyone. We should all, testers especially, remember to view things from the other side.

After a few months of trying to initialize a response, It has now been three weeks since we first heard back from DJI and they sent out the beta. That is plenty of time to get the fix implemented and released to the public if they were working diligently on it. It is clear they are not.... at least not as fast as we should be demanding to accomplish our original goal!

I originally posted that we should ramp up the Project again on the 19th, but as Ianwood stated, there should be a second beta sent out this week. We should give them one more week, but after that come on.... It would be hard for anyone to argue that it is not getting ridiculous at that point.

Your "perception" observations are probably astute, and just to be clear - I completely agree that it is appropriate to keep up the pressure - my point was that I'm unconvinced that releasing the beta is appropriate. Why do I think there would be a landslide of complaints? Because even if it is no worse than the present firmware, and even if it fixes the problem in some cases, the continued problems will lead, as always happens, to accusations of releasing half-assed firmware that should still be in beta.

As for how long it should take - I'm not defending DJI and I'm on record saying that I do not understand, based on my conclusions on the nature of the problem, why there is not a simple and quick fix already implemented. But, if it turns out that it isn't simple, then claiming that it should already be fixed is just guesswork.

You are right. It is guesswork and I shouldn't assume that it is an easy fix and that three weeks should be enough time to push out a fix, but I think my overall point is just that: We shouldn't assume that just because DJI has responded with a beta that everything is fine and a fix is right around the corner and anyone who continues to complain about the problem or the time it is taking should be told to hush it. (I'm not saying people have been told that, just that it seems to come across that way sometimes)

My main argument, and it seems as though we are on the same page here, is that Project Drunken Sparrow, whose simple purpose is a working fix to the problem, should not stop fighting to reach its goal until it is attained. Our only way to fight is through pressure. It seems as though the fight has stopped in those who have been 'appeased' with the beta.

 

[ianwood]

Thank you ElGuano. Succinctly put. I didn't even want to try to go there for fear I would write a whole page on the complexities of release management.

Everyone, please read this carefully
I am not sure everyone is clear on the strategy I set out when we started this whole thing months ago:

• Collect information.
• Establish a working relationship with DJI.
• Get them to fix the problem.

...and if DJI doesn't demonstrate the right level of effort, enter the sparrow.

• Launch social media campaigns until DJI takes the problem seriously.
• When DJI responds and shows they are working on it, the sparrow stops the campaign, observes DJI closely and helps.
• If DJI loses focus or doesn't respond, we bring back the social media campaigns even stronger than before.

The fundamental point is if you are trying to motivate someone, you have to establish a pattern of incentive. You punish them when they misbehave. You reward them when they behave. You do not punish them when they are trying to do the right thing.

Yet some of you want to do just that. You want to continue slamming DJI with media campaigns while they are actually working on more beta firmwares for us to test. That breaks the whole incentive strategy. I cannot stop you but you would be putting this project at risk by doing that.

 

[landonkk]

Thank you ElGuano. Succinctly put. I didn't even want to try to go there for fear I would write a whole page on the complexities of release management.

Everyone, please read this carefully
I am not sure everyone is clear on the strategy I set out when we started this whole thing months ago:

• Collect information.
• Establish a working relationship with DJI.
• Get them to fix the problem.

...and if DJI doesn't demonstrate the right level of effort, enter the sparrow.

• Launch social media campaigns until DJI takes the problem seriously.
• When DJI responds and shows they are working on it, the sparrow stops the campaign, observes DJI closely and helps.
• If DJI loses focus or doesn't respond, we bring back the social media campaigns even stronger than before.

The fundamental point is if you are trying to motivate someone, you have to establish a pattern of incentive. You punish them when they misbehave. You reward them when they behave. You do not punish them when they are trying to do the right thing.

Yet some of you want to do just that. You want to continue slamming DJI with media campaigns while they are actually working on more beta firmwares for us to test. That breaks the whole incentive strategy. I cannot stop you but you would be putting this project at risk by doing that.

OK, I get it. I'll stop raising the issue and let it play out. However a few points:

1) Those may have been your plans all along but it was never laid out as such. In your post 'Project Drunken Sparrow' (viewtopic.php?f=19&t=16598), your stated goal was to exert pressure on DJI to get a fix, NOT exert pressure to get a working relationship or exert pressure to get a beta. I am simply pushing to accomplish your stated goal.

2) Once you recruit people to join your cause, you have to allow that cause to become the people's cause. It is no longer your own.

3) I really fail to see how continued pressure would have a regressive effect. DJI is going to stop working on a fix to an obvious problem that is causing them negative press and a hit to their reputation because we didn't play nice?

All of this said - I don't want to sound like I am unappreciative or think that this is all a failure. I just want the project to succeed in its stated goal and have a slightly differing view on how to get there.

 

[SilentAV8R]

It is interesting how the tone has changed. Back in May I was taken to task for saying this may be more complicated than many were thinking. In fact, this is what I was told:

You make it sound like adjusting for declination is some sort of next generation rocket science. It's literally 3 lines of code and a table.

Turns out it may be a bit more complicated than that. Might not be rocket science, but apparently it is more than "3 lines of code and a table" as well.

 

[Raffelos]

Excellent statement, after the beta testing, DJI has a new advocate.

Sent from my SM-N900T using Tapatalk

 

[sar104]

It is interesting how the tone has changed. Back in May I was taken to task for saying this may be more complicated than many were thinking. In fact, this is what I was told:

You make it sound like adjusting for declination is some sort of next generation rocket science. It's literally 3 lines of code and a table.

Turns out it may be a bit more complicated than that. Might not be rocket science, but apparently it is more than "3 lines of code and a table" as well.

Or possibly, and this is my concern, is that it is "3 lines of code and a table", but instead they are trying to fix it with rocket science.

 

[Andy T]

the new logo of the forum by chance would be an analogy to the problem j-hook and TBE of phantom? :lol: :lol: :lol: :lol: :lol:

LOL love it :lol:

 

[BigBadFun]

-----------------------

 

[ianwood]

It is interesting how the tone has changed. Back in May I was taken to task for saying this may be more complicated than many were thinking. In fact, this is what I was told:

You make it sound like adjusting for declination is some sort of next generation rocket science. It's literally 3 lines of code and a table.

Turns out it may be a bit more complicated than that. Might not be rocket science, but apparently it is more than "3 lines of code and a table" as well.

Yeah, that was a pretty harsh response from me. Even worse I used "literally" were it does not belong. But I still think it should be simple addition in a well designed system. The Naza doesn't look so clean. For all we know, it might be a heaping pile of ugly hacks that makes simple fixes like this really hard. Maybe the MC doesn't have any memory left for even a simple lookup table. No one here can know for sure.

And even if DJI could implement the fix in a few lines of code and a small table, there are typically a lot of steps that need to be completed between coding the fix and getting it into everyone's hands as ElGuano described well. I doubt DJI's process is very sophisticated given the number of production defects but even a simple release process can take weeks to complete.

And Landon, the strategy is in this thread, maybe not in complete detail in one post but it's in this thread multiple times. The one thing I've asked with the sparrow campaign is we do it together in a planned surgical strike. That's what makes it powerful.

The hard part is knowing when to do it. No one here knows how much DJI is working on this. They could be feverishly slaving away on it or they could have one 2nd level engineer working on it while he also does 10 other things. It's likely something in between but no one here knows (and please don't try guessing). I can only go on what I get from them and as I understand it, they are putting together a new beta for us.

If that doesn't materialize or generally if after a reasonable amount of time, there is no sizable progress, the sparrow goes back into attack mode and I'll be pushing all you guys to post on their social media like crazy. What we do next and when is not my dictate, but if we want to be effective, each sparrow strike needs to be timed well and thought out. Scorch the earth too many times, DJI may change their mind about fixing it.

I don't know what else to say at this point. I am really not enjoying the allegations of being a DJI stooge or resting on my laurels because I have a beta that kind of works. I'm not that cheap. When DJI sends me a fleet of S1000s, maybe I'll be their stooge. Until then, this problem is still not fixed.

 

[SilentAV8R]

I'm familiar with the level of effort it takes to get software changes incorporated into hobby electronics. It is never as simple as it might seem to those of us not actually doing it. Especially when dealing with a language barrier. But progress seems to be being made and I still find mine useful to fly, if perhaps a little heavier load on the pilot. Sort of a semi-automatic camera versus a full point and shoot.

 

[Pazz]

HI Guys, I have been reading with interest regarding this problem. I currently fly a vision 2 & operate in south east Australia. Mag dev 22 degree.
I have experienced the issues you refer to, but only randomly. I fly on average 3 times a week, with 4_5 missions per flights from various locations each time.
While I respect your views on this matter, suspect the problem is not the compass, but rather the GPS system in general. It is not a stable system & has many
inaccuracies in minute to minute operation. At best it's reliable accuracy is 5- 30 meters &is influenced by location, sattilite positions,atmospheric weather conditions, local magnetic variation & constant updates from the GPS system..
My guess is that the vision auto pilot system struggles to keep up with GPS updates as it is in flight & understandably gets confused as to where it is, hence the TBE. Also explains hooking as the compass bearing & GPS don"t always agree.
I also explains why it is a random event which I experience, as most of the time it flies fine, provided I calibrate the compass every flight in each location.
It appears to be an issue across all NAZA type systems in what ever brand you name which leads me to conclude that it the GPS system thats the problem & this will never be resolved by the drone makers alone.
The best solutions will be the learn to fly them either in manual Or atti mode & not rely on GPS solely. Geoff

 

[riblit]

It's not the GPS system Pazz, it's the magnetic declination. That's why it's only seen in areas of high magnetic declination. I'm now flying the beta firmware and can tell you the TBE is almost completely gone and the system is learning faster. There are still some problems with course lock to be sorted. If it was the GPS system it would be there irrespective of magnetic declination.

 

[obiwan_pierogi]

It's not the GPS system Pazz, it's the magnetic declination. That's why it's only seen in areas of high magnetic declination. I'm now flying the beta firmware and can tell you the TBE is almost completely gone and the system is learning faster. There are still some problems with course lock to be sorted. If it was the GPS system it would be there irrespective of magnetic declination.

Not to mention, if you're positive declination, like I am (17.3+), the rotation is CC, and left J hook.. I understand it's the opposite if one has a neg. declination.

 

[xtonex]

It's not the GPS system Pazz, it's the magnetic declination. That's why it's only seen in areas of high magnetic declination. I'm now flying the beta firmware and can tell you the TBE is almost completely gone and the system is learning faster. There are still some problems with course lock to be sorted. If it was the GPS system it would be there irrespective of magnetic declination.

Not to mention, if you're positive declination, like I am (17.3+), the rotation is CC, and left J hook.. I understand it's the opposite if one has a neg. declination.

Correct I'm at -11 and my hooks are to the right

 

[BigBadFun]

At best it's reliable accuracy is 5- 30 meters & is influenced by location...

Woah.... I'd be returning that immediately to DJI if that was the case?

 

[Doono]

HI Guys, I have been reading with interest regarding this problem. I currently fly a vision 2 & operate in south east Australia. Mag dev 22 degree. Geoff

What's a "Mag dev 22 degree"?

 

[droneranger]

Pazz is referring a to magnetic deviation of 22 degrees (however, he does not indicate whether it is a + or - value). Magnetic deviation is a local magnetic error produced by nearby iron objects or ferrous deposits in the earth itself. It is expressed in minus or plus degrees, either (West or East).

 

[droneranger]

HI Guys, I have been reading with interest regarding this problem. I currently fly a vision 2 & operate in south east Australia. Mag dev 22 degree. Geoff

What's a "Mag dev 22 degree"?

Pazz is referring a to magnetic deviation of 22 degrees (however, he does not indicate whether it is a + or - value). Magnetic deviation is a local magnetic error produced by nearby iron objects or ferrous deposits in the earth itself. It is expressed in minus or plus degrees, either (West or East).

 

[N017RW]

HI Guys, I have been reading with interest regarding this problem. I currently fly a vision 2 & operate in south east Australia. Mag dev 22 degree. Geoff

What's a "Mag dev 22 degree"?

Pazz is referring a to magnetic deviation of 22 degrees (however, he does not indicate whether it is a + or - value). Magnetic deviation is a local magnetic error produced by nearby iron objects or ferrous deposits in the earth itself. It is expressed in minus or plus degrees, either (West or East).

It's best described by the difference between Magnetic North and True, or Geographic, North [pole]. (Geo)Magnetic North actually shifts over time and even reverses!