Pixhawk Feedback?

[R_Lefebvre]

So, just bumping this up again. Previously Dazzab had asked why nobody is using Arducopter/Pixhawk for professional AP, and I said there were, but I couldn't show examples. This is one of the examples I was talking about, I finally found the public release. Skip ahead to 3:55. This was actually done with nothing more than a GoPro on a Tarot 2-axis gimbal flown on an Iris. The footage isn't the most spectacular thing ever, but it's pretty good considering the system being used. Not bad for a 2kg back-pack-able quadcopter.

 

[dazzab]

So, just bumping this up again. Previously Dazzab had asked why nobody is using Arducopter/Pixhawk for professional AP, and I said there were, but I couldn't show examples. This is one of the examples I was talking about, I finally found the public release. Skip ahead to 3:55. This was actually done with nothing more than a GoPro on a Tarot 2-axis gimbal flown on an Iris. The footage isn't the most spectacular thing ever, but it's pretty good considering the system being used. Not bad for a 2kg back-pack-able quadcopter.

That is nice. Great to see someone using something other than a Phantom to lift the GoPro. But when I said professional AP I was referring to heavy lift copters. Show us something done with a full size DSLR or a Red Epic being flown using a Pixhawk FC.

I have no doubt that the Pixhawk can fly smaller copters quite well. It's the big beasts I'm speaking of. Let's take a very recent incident as an example. While beta testing the next release of Arducopter a few copters simply fell from the sky. It was discovered that the landing code was falsely identifying that the copter had landed when it was actually still at altitude and shutting down the motors. Yep, copter in the air, copter crashes to ground. It doesn't appear that fault ever showed up in the previous production software however Randy stated that he wasn't sure why. Of course, that bug was fixed in the beta so the problem won't exist in the next release so all is well.

The point of that example is what I've been saying all along. Arducopter is fantastic, no doubt about it. But it is still in development. No one in their right mind is going to put a $20,000 copter in the air with development software. Well, I put a $8,000 one in the air and I paid the price. That was my choice and my contribution. For now, I stick with the FC that all the pros use. I have no doubt that the day will come when I will be flying with a Pixhawk running Arducopter. But that day is not now.

 

[jfro]

Well, I put a $8,000 one in the air and I paid the price. That was my choice and my contribution. For now, I stick with the FC that all the pros use. I have no doubt that the day will come when I will be flying with a Pixhawk running Arducopter. But that day is not now.

dazzab, curious as to how much of the $8000 your were able to recover and re-use....

I had a pixhawk fall out of the air with go pro and it cost me around $400 dollars. Motors cut out because something expectantly happened after a auto flight after attemping to land manually. In changing modes on a 2 combo switch, I went through a mode that when I cut back on the throttle, it shut them off and when I raised throttle it was to late, it had shut down.... My fault, seems crazy that there is a cutoff non recoverable mode in any mode.

GPS pulled the connector off the circuit board, so I had to buy a new pixhawk. First time ever having to replace a flight controller.

Anyway, I am back flying and testing, but it will be a long time before I consider putting it on a large rig. If I do, it will get 30 + flights with a weight under it and limit the modes I fly in... Which is not why we buy the Pixhawk. We buy it cause it can do so much.... Conundrum.

 

[R_Lefebvre]

That is nice. Great to see someone using something other than a Phantom to lift the GoPro. But when I said professional AP I was referring to heavy lift copters. Show us something done with a full size DSLR or a Red Epic being flown using a Pixhawk FC.

I have no doubt that the Pixhawk can fly smaller copters quite well. It's the big beasts I'm speaking of. Let's take a very recent incident as an example. While beta testing the next release of Arducopter a few copters simply fell from the sky. It was discovered that the landing code was falsely identifying that the copter had landed when it was actually still at altitude and shutting down the motors. Yep, copter in the air, copter crashes to ground. It doesn't appear that fault ever showed up in the previous production software however Randy stated that he wasn't sure why. Of course, that bug was fixed in the beta so the problem won't exist in the next release so all is well.

The point of that example is what I've been saying all along. Arducopter is fantastic, no doubt about it. But it is still in development. No one in their right mind is going to put a $20,000 copter in the air with development software. Well, I put a $8,000 one in the air and I paid the price. That was my choice and my contribution. For now, I stick with the FC that all the pros use. I have no doubt that the day will come when I will be flying with a Pixhawk running Arducopter. But that day is not now.

All autopilot software is under development. We're just open about it. You KNOW when you're getting a beta firmware, and choose to fly that at your own known risk. Unlike some of the other big guys, who put out a "stable" release that seems to have not been tested at all, then a week later, change it an re-release with the same firmware number.... just to not admit fault, and leaving users to wonder exactly which version of the code they have...

You're kidding yourself if you think the others are not under development. I mean seriously? Even SuperX changes the firmware every so often.

ie: http://xaircraftamerica.com/products/superx-flight-controller

Now on firmware version 1.08 as of 12/2/13

So guess what, that would seem to be the 8th release of "production" firmware. (usually 1.0 is used to signify first production release)

And you know what, next year, when they release a brand new flight controller, with brand new firmware, that would probably be an evolution of 1.0. You could probably call it the 2.0 firmware, but they will of course call it "SuperY 1.0 Firmware".

We just don't force our users to buy new hardware every time we have a major software release. The two things are uncoupled. But that does not mean that we never have a stable release.

And Randy darn well does know why that crash happened, and it was not a risk with the previous software. For 3.2, he tried to relax the landing checks, because some people complained that the copter spent too long on the ground with motors running during an auto-land. It didn't work, so he went back to the more rigorous but safe setting.

As far as big machines flying on it, I'll keep looking for an example I can share. I used 3.2 rc9 to fly my big 800 gasser heli. It flew very well. I also use it on a 16lb camera-ship heli.

Also, you may be aware that Tridge just won the Outback Challenge, with non-stable-release Arduplane software. For those not aware, it involved flying a large gas airplane, completely autonomously from take-off to landing, flew a few miles out to the search area, lawn-mowering the search area. All the while, there's a camera system scanning the ground, with autonomous imagine recognition software looking for a dummy on the ground. And after locating the dummy over a search area of several square miles, dropping a water bottle which landed within a few meters of the dummy.

All with "not stable software".

My fault, seems crazy that there is a cutoff non recoverable mode in any mode.

I'd need more info. I cannot think of any mode where there would be an unrecoverable throttle cut mode. If there is, it's a bug we need to fix. My suspicion, is that you had the throttle stick on the bottom, and you flicked through Stabilize mode. Stabilize does stop the motors if the stick is all the way down. But they also restart instantly if you raise it. I imagine you were close to the ground, and didn't have time to respond. But this is user error, not a system failure.

 

[dazzab]

Also, you may be aware that Tridge just won the Outback Challenge, with non-stable-release Arduplane software.
All with "not stable software".

Actually, Canberra UAV won the outback challenge, not Tridge. I'm a member of CUAV and have known Tridge for quite some time. I do a lot of photography and video for the team. Several other teams using Pixhawk did _not_ win the contest. So what's your point? Comparing Tridge using beta software to anyone else on the planet is just plain silly. Readers might want to read Tridge's summary of that flight as it actually mentions a very serious issue with the software that caused a problem at takeoff. Point taken?

The Outback Challenge was won using Ardupilot which Tridge has contributed heavily to along with many others - not Arducopter. Tridge would be the first to say he would never had accomplished that feat if it were not the great team that came together. It took a lot more than a Pixhawk to accomplish that. If Tridge ever becomes the lead developer of Arducopter then I'm in. And that's not to take away from Randy and the others, including you. You've done a great job and I really look forward to the day with Arducopter works well enough to meet my needs. BTW, I actually use Pixhawk on planes and a couple of smaller copters so it's not like I'm anti Pixhawk or something. I've probably spent many thousands of dollars with 3DR and spent a lot of time with the FOSS projects.

You don't agree with or respect my opinion. We get it. If you are comfortable running Arducopter/Pixhawk on expensive gear then go for it. I am not and I know many others who are not. Your credibility weakens when you argue with those out there doing the job. You should be spending your time working _with us_ if you want better results.

 

[dazzab]

dazzab, curious as to how much of the $8000 your were able to recover and re-use....

I had a few crashes over the 9 or 10 months I tested Arducopter on my SkyJib. In comparison and almost as a control case, I was flying with a friend with a similar copter that has never crashed and is extremely stable. I started before the Pixhawk was around. It probably cost me about $1,000 in parts which is actually fairly cheap with Droidworx gear. But the real cost was time and frustration. I had the same issue you mentioned of problems when switching modes. Twice I switched it out of RTL to stop the auto landing and take control, only to have it fall from the sky. Some of the issues I had were certainly my responsibility and others don't make sense to me. Look up my user name on DIYDrones or here if you want more info/details.

I've moved on from all that and I now concentrate on how to obtain professional level imagery. I have a Wookong on the SkyJib and use SuperX on my hexa and quad. I spend virtually no time at all dealing with the FC other than to check logs from time to time. The copters fly quite well. My issues now are with gimbals, cameras and video Tx. But I'm getting there and progressing nicely. Net week I'm even doing training for my Controllers Certificate from CASA (our FAA) which is part of the process of being involved in commercial activities.

 

[jfro]

My crash turns out was operator error, as I mentioned earlier..... I have had similar issues with other FC when I cut throttle real quick and then brought it back when I panic. Turns out changing modes with two switches and then having the motors just flat stop were too much for my brain too handle. I think my problem is there is a lot to learn and REMEMBER. Maybe not the best environment for me and my brain. I'm still working on it and it may be that I have to stay away from certain features. I still have a like/"don't like so much" relationship with it.

I've moved on from all that and I now concentrate on how to obtain professional level imagery. I have a Wookong on the SkyJib and use SuperX on my hexa and quad. I spend virtually no time at all dealing with the FC other than to check logs from time to time. The copters fly quite well. My issues now are with gimbals, cameras and video Tx. But I'm getting there and progressing nicely..

As to your comment on working with "gimbals, camera, and video Tx", I would agree that's what I'm focusing on (as well as learning to baby my batteries to last longer). For myself, I'm about ready to take the gimbal off the list. I'm getting some pretty good footage off the new Phobotic Centerpiece controller. Testing on the new beta software that is imminent is another step up the ladder. I've taken the Movi 5 off my wish list and will now build a 2nd gimbal with the CP. It's not perfect, but good enough and hopefully over the next 6 months, some fine tuning will make it even better.

Camera's are already better than my eyes in almost every way except for dynamic range, so I'm content there. Last couple issues for me is an affordable HD video tx, becoming a better pilot, and reliability of the gear in the air.

 

[R_Lefebvre]

You don't agree with or respect my opinion. We get it. If you are comfortable running Arducopter/Pixhawk on expensive gear then go for it. I am not and I know many others who are not. Your credibility weakens when you argue with those out there doing the job. You should be spending your time working _with us_ if you want better results.

The reason I don't respect your opinion here, is because your argument is not reasonable. It's seriously flawed right from the start, and you refuse to acknowledge the reasons for that. Here they are again in point form.

1) You state you will not use Arducopter for professional purposes until it goes "stable".

2) Arducopter will never go stable. It will never stop moving.

3) No other software is "stable". They all go through a series of releases and bug fixes, etc. Even the really really expensive stuff.

 

[dazzab]

1) You state you will not use Arducopter for professional purposes until it goes "stable".

I actually said under development. There is a difference.

2) Arducopter will never go stable. It will never stop moving.

That's an issue for me. Too many developers taking it too many directions. The other flight controllers are less complicated and seem to have locked things in better for the needs of aerial photography. You yourself once identified that the APM was running out of capacity for operating octocopters but the developers didn't know it until the users told you. On several occasions users have had to identify issues even after being told, like me, that they were wrong.

Of course the commercial products go through revisions. If they are so bad, I wonder why they have so much of the market I'm discussing? I guess all the users are just stupid, right? Thank goodness you came along to educate all the operators here actually making a living doing this professionally. Yet you still haven't shown any evidence in the form of a commercial operator flying professional equipment using Arducopter.

3) No other software is "stable". They all go through a series of releases and bug fixes, etc. Even the really really expensive stuff.

I've been in IT for decades. Stop insulting my intelligence. There's a big difference between the processes used to release commercial software for a specific use as compared to open source software to meet many needs. Take a look at the Phantom. Of course it works great. It's got a FC in it tuned specifically for the copter it's in. I'm sure even you could do that.

Feel free to keep attacking me personally. I'm sure other users here can see through it. You would be naturally biased. You leave out important information and perspectives such as you did today when attempting to use the success of CUAV at the Outback Challenge as an example. You show little respect for others and their reported experiences.

You appear to want to just preach how great your software is rather than have a real conversation. I really don't understand why you are even here. Anyone who is interested in the Pixhawk and Arducopter really should be getting information from DIYDrones rather than one developer with an obvious chip on his shoulder.

If people want to hear about my experiences with Pixhawk and Arducopter I would appreciate it if you would just STFU and let us get on with sharing and assisting each other.

 

[R_Lefebvre]

I actually said under development. There is a difference.

No, I don't think there is. What exactly is the difference you are suggesting? Is MS Windows still under development? They release patches almost weekly. Oh, I guess Windows XP is now no longer under development. No more patches for it. Is it better to run Windows XP since there won't be any more updates? Or is it better to run Windows 7, which still receives updates (and bug fixes?).

Do you know when was the last time we had to do a "patch" on the 3.1 branch of code? May 26, 2014 is when 3.1.5 was released. So 4 1/2 months since a bug has been identified. Not too bad. How many patches has DJI or SuperX had in that time? The 3.2 branch is a major release. IMO, it should actually be called 4.0, it's so big. The only reason it's not is because it still runs on APM, and they are reserving 4.0 for the first release not supported on 8-bit. Which, is a sticky situation since the best things about 3.2 don't run on APM anyway. Anyway, 3.2 is basically a "new product", and resets the bug fix counter.

A better question is, how many bugs are they NOT fixing? Would you know? Can you talk to the developers? Do you have access to data logs sufficiently thorough that you would even be able to identify software bugs if there were any?

I categorically reject this idea that because we are open and honest about our bug fixes, that somehow that makes us less reliable than a closed source system where bugs are not acknowledged and fixed, or hidden from release notes. And I refuse to accept that since we will not stop moving forward, that that somehow makes us less reliable.

That's an issue for me. Too many developers taking it too many directions. The other flight controllers are less complicated and seem to have locked things in better for the needs of aerial photography.

You still have not identified what exactly it is that makes a flight controller better for aerial photography. The only thing I've ever been able to figure out that you want, is "simple to use through lack of features". Yes, as has been stated before, the boat has sailed on that one, you'll never get that with Arducopter. You can't say that what you're looking for is reliability, because you have no scientific evidence to show that any other system is more or less reliable than any other. Nobody has that. All we have is perception. And perception is, at least from you, that Arducopter is less reliable. My opinion is that is simply based on the fact that we have more user error due to complexity. And an open and honest bug fixing process. Again, neither of those two things will ever change.

Here again, you cast aspersions to this idea that Arducopter can't fly a large machine. You may have had trouble doing so, but others aren't. I've helped get it working on a 26" prop machine. And I got it working on an 800 size gas helicopter. Pretty big machines. So it is not a valid statement that it doesn't work.

You yourself once identified that the APM was running out of capacity for operating octocopters but the developers didn't know it until the users told you. On several occasions users have had to identify issues even after being told, like me, that they were wrong.

No, actually, that's not what happened. I myself identified the problem. However, do users sometimes find problems? Yes, absolutely. That is the beauty of open source. Again, the fact that we allow users to help find bugs, and then actually fix them, and acknowledge the fix in the release notes, is a strength not a weakness.

I've been in IT for decades. Stop insulting my intelligence. There's a big difference between the processes used to release commercial software for a specific use as compared to open source software to meet many needs. Take a look at the Phantom. Of course it works great. It's got a FC in it tuned specifically for the copter it's in. I'm sure even you could do that.

Oh please, don't feign insult, I would have no way of knowing what your background is. All I can do is look at the statements you make. Like this one. So you're suggesting the Phantom it has commercially released software that went through a process. Now lets have a look:

1. Main controller firmware updated to v3.06.
2. Central board firmware updated to v1.0.1.32.
3. DJI Vision App updated to iOS v1.0.42 and Android v1.0.52.
4. Phantom 2 PC Assistant updated to v3.2

So, it appears to have gone through 6 main controller firmware updates since release. And that is despite the fact that it is actually the second generation of that product, where the first generation went through what, 20+ releases? Huh. So much for a thorough commercial release. And the revision numbers would seem to indicate some number of central board firmware updates, DJI Vision Ap updates, and PC Assistant updates.

So how or why is this better, exactly?

Feel free to keep attacking me personally.

I have NEVER attacked you personally. I am attacking your arguments. This is a debate. YOUR statements about "chip on the shoulder" and judgement about my respect for others ARE personal attacks. And they don't warrant further comment.

If people want to hear about my experiences with Pixhawk and Arducopter I would appreciate it if you would just STFU and let us get on with sharing and assisting each other.

Actually, I think people would prefer to hear less of your experiences. You don't like Arducopter. You never will. Despite your suggestion that someday maybe you will, I can tell you right now, you won't. Because again, we will never stop having an open and honest bug fix system. And we will never stop innovating and developing. So there really is no point you coming in here, and constantly telling us how you don't like the system. Yes, we get it. And that's fine. It's not for everyone.

Most forums actually do appreciate having somebody around who can actually comment about the system from an insider's perspective. Not a user with sour grapes who has nothing more to share but the same story of how he couldn't get it working for him.

 

[dazzab]

Thanks for your time Rob. Some interesting points in there. I guess at the end of the day we have to agree to disagree. It's a shame we are so far apart as I'd love to have these conversations over coffee and more importantly to fly together. I am quite intrigued by your traditional helis which are very cool. At this point I'd really like to hear from others using Pixhawk and Arducopter on heavy lifters. We've both made our points and I don't see any value in just endlessly quoting and debating.

 

[jfro]

We've both made our points and I don't see any value in just endlessly quoting and debating.

I think that's a good idea. Let's move on and talk about our experiences with Pixhawk, whatever they may be.

 

[jdennings]

Here’s an experience/opinion/comparison, fwiw ... I currently fly two heavy lifter octos for filming, right at the junction between semi-pro and pro. I’ve been at it for several years, flying both apm then pixhawk, along with DJI wkm and Naza. I also fly a half dozen smaller copters including a Phantom 2 with H3d3, some FPV copters, and a few “test” or backup copters.

In the past year or so I’ve seen Pixhawk evolve from a very good flight controller to the best, imho (only comparing only to DJI naza and wkm), in terms of flight quality and reliability.

As far as flight quality (smoothness, precision), I used to have a slight preference for WKM prior to AC 3.1, then about equal with current stable release of AC, then slight preference for Pixhawk with AC 3.2 (still in beta).

In terms of reliability, I ‘ve been feeling much more confortable with AC for at least a year if not more. Reason is fairly simple: I know AC is being tested extremely thoroughly, and that any issue is not only identified but scrutinized by an army of very talented people, and dealt with, openly and transparently. If the specter of a unexplained crash or flyaway comes up, it gets to the top of DIYD in a heartbeat, gets discussed, analyzed to death, including down to the code level if need be, and always explained and dealt with. In addition and just as important, I have yet to see a report of a situation where switching to manual would not have saved the copter. With DJI, well, it’s a black hole. While I am fairly certain that many flyaways are due to the inexperience of their users, I have failed (and believe me I’ve tried) to dismiss reports of unexplained crashes or flyaways, or situations where switching to manual was innefective. This includes events involving very experienced and some of the very best film flyers. It actually worries the heck out of me when flying even my lil’ phantom, and I still feel a tad unconfortable when flying wkm. Meanwhile as far as A2, which after all is DJI’s flagship and should be better than wkm, I won’t touch it with a 10 foot pole. I’ve seen too many reports of Red Dragons going down, and when I see a guy like Kloner retiring it, well, ‘nough said ... And again, flying an FC from a company whose flagship product is rejected by some of the best doesn’t exactly contribte positively to my comofrt level. SuperX, which appears to be the third and only alternative, I have no experience with. It sounds very good yet, again, it’s closed and support appears to be minimal. Edit: Forgot Mikrocopter. But I lost them when they came up with the license thingie for waypoints and commercial use quite a long ago...

Another way I look at this, on the topic of reliability, and this touches on the topic of open vs closed source or designs, a topic I’ve dealt with for many years in the field of computer security and cryptography (unpublished crypto algorithms should be avoided at all cost): Suppose I am about to embark in a long cross country car race, and I have the choice between two cars. The first is one that is regularly inspected by hundreds of mechanics or “advanced” drivers, some of which not hesitating to pull it completely apart, rebuild it, change components, push it to unheard of limits, inspect parts with microscopes, etc ... The other one is nice and shiny, drives nicely, it’s got a great reputation and is known to work well yet the hood is locked, no one has peaked inside, the only ones familiar with what’s beneath are a team of very talented yet silent designers and mechanics in China. Both cars have had failures. Yet the first has successfully been used in races much more difficult and taxing than the one I plan to be in. The second has been very good, yet has failed in a number of instances with no post-mortem explanation from their drivers or their makers. Both are very complex machines, the first with regularly come up with new versions, the first with many new features, that I may or may not need for the race, yet features not available for the second. The first has multiple built-in redundancies, the second does not. Which one would I choose to race if reliability if my #1 concern? No brainer as far as I am concerned.

As far as ease of set-up, Pixhawk still lags and DJI or SuperX still win. It’s not so much an issue for me as I’ve invested enough time with Ardupilot that it takes me no more time to set it up than DJI. But this would certainly be different for someone new to Ardupilot, and if you want to get up in the air as fast as possible, have a “standard camera”, want no tweaking, an S1000 and Zenmuse is one alternative. On that subject I’ve got to say that even with experience, I was still blown away when I got my Phantom 2 with gimbal, and was flying perfectly acceptable footage with 15 minutes of unboxing. (Really... Batteries where 2/3 full, a few minutes easy calibration, and off I went, I couln’t wait ...).

Yet the price to pay for an all in one, easy to set-up ready to fly solution is lock-in. With Pixhawk, I have a flexible system and am insured against the future. New better GPS comes up, centimeter precision atti/alt hold with Lidar? No problem, easy swaps. Faster CPU, running on Linux, comes up? No problem, will just replace the board, it will configure and run the same. Want to add aerial mapping to my services? No problem, easy auto grid is already here. Need some auxiliary servo work, or better telemetry than stock? Got it. Want to film with follow-me? No problem. Need to tweak fast with auto-tune for those new motors and 20” props? Easy. With my DJI FCs I can’t do any of this. I am locked to a black box. I can’t replace a Gopro on a phantom with a similar small camera. I can’t replace that Gh4 with another camera with heavy lens, even for that one time only shoot, on that Zenmuse. All this may not be a DJI deal breaker if all I want to do is filming, don’t have special requirements, and my current cameras are here to stay. But it does severely constrain, and considerably adds up cost-wise, especially on long periods of time.

So in short, I find my wkm gathering more and more dust, and it's a joy to fly Pixhawk and I can't think of anything lacking. No question the Ardupilot time investment has been significant, in fact I sometimes find myself wondering about all that time spent that could have been used to “just get out there, fly and film”, the only thing that matters, after all. The same happens when I find myself building custom parts for that new gimbal attachment, vs just buying a Zenmuse and an S1000, just spend $8-10k and be done with it. But then just around that time the reliability factor kicks in, along with the flexibillity and future proof argument. Not to mention that understanding what’s under the hood does much more than feeding the addicted geek in me: it simply makes me a safer flyer.

Oops, meant to just quickly write a paragraph or two ... Time to get back to “real” work ..

 

[jfro]

jdennings;

Thanks for your insight. I'm a bit like you from what you described, only I haven't gotten over my fears to put my Pixhawk on a bigger MR.

Two changes I'd have to figure out is the 6s battery, do I want to just send 5v to the rails or do I want to go find or build a 3rd party power adapter (which seems not something I really want to build). The other is whether or not auto tune works with my 400kv 3520 or 3515 motors. I've read that it sometimes doesn't work with the lower KV motors. Have you had success with autotune?

On my small quad, I was relatively happy with my tuning, but after switching to autotune, I thought it was a bit better and a whole lot quicker.

Again, thank you for your opinion piece, I, for one, appreciate it.

 

[R_Lefebvre]

I'm just working on the next feature, will probably be in 3.3: Collision avoidance. I've got a Lidar aimed out the front.

 

[Holger]

Well we had Ardu 2.6 on a Cinestar frame last year and it worked well..

 

[jdennings]

I've been cautious with APM 2.x on octos and hexes, actually I've pretty much retired them. Although it's definitey supported and the dev team has been working hard on optimizing every bit of the code to fit on 8-bit MC boards, it is on its last legs space wise and while it will work you might not get all the performance you could get with the 32 bit Pixhawk ...

 

[R_Lefebvre]

Yeah, I really don't see why anybody would use an APM anymore, except for I guess all the hobbyists who bought one and can't afford to just buy a Pixhawk. But if you're doing anything kind of serious with the system, it doesn't make sense to not upgrade.

 

[R_Lefebvre]

Arducopter used on a mid-size camera ship:

 

[SAVIO]

Pixhawk is a nice 32bit FC, but those DF13 conectors are wayyy too fragile, i myself broke the GPS conector and the only way i found so far is to buy a new one. I am wondering the French DROPIX has better connectors for the same archtecture. Thats just my opinion.
I apologise my english skills.
I am brazilian.