To draw a comparison with wedding photography, the best photographers still make a very good living at it even though 'anybody can take pictures'. If you want it done reliably and care about quality you pay for a professional. Or you can take your chances with Uncle Bob and his fancy Nikon with the big long lens.
Even this isn't so easy. At our wedding, we used a guy who ran a very small shop from a rural area, for all I know he had a "day job" working in a factory. He was very affordable. I was worried we'd get some podunk results, but he did a great job. My brother, used a high-priced professional from a big studio in the city. Paid 3-4X as much as us. The results were pretty bad. In fact, their #1 photo, the one they have framed, is a setting that *I* pointed out and actually had to push for the photographer to use. Most of the other photos are unusable.
It's just that it's focus was never on what was best for aerial photography. It's not set up to be mushy as Robert pointed out because the developers never understood what a photographer would need IMHO.
It's this part that of the discussion that I really struggle with.
I've asked many times what it is that photographers want/need, and nobody can ever provide a descriptive list. What exactly is it that Arducopter does not do well for you? Reading between the lines, the only technical thing I've been able to figure is basically, soft or mushy control. Fast is actually, we used to have softer control. It wasn't until Autotune came out that we ended up with razor sharp responses. But you can still make it softer if you tune it right. Regardless, with 3.2, you'll be able to set it up to be very soft to inputs, yet still have razor sharp stability.
DJI, do still have an edge on the performance in GPS Hold. I'll give them that. I can't figure out how they hold so well, without being "busy" like Arducopter tends to be.
But other than that... the only other thing I can see is simply easy of use. A controller designed to do only one thing, is easier to use if that is what you want to do. Arducopter does have way more features and settings, because it can do so much more.
But there's another thing... cost. It's weird, but there's a lot of people out there who won't consider it because it's simply too low cost. So much of this market is about male jewellery. "You get what you pay for". Fancy cases and high price tags make people think the system is better. Helicopter guys won't even touch it cause it's too cheap. Somebody flying a $10,000 camera, is more comfortable with a $2000 controller than a $300 controller. And their clients are too. But this isn't a rational decision making process.
It's a project in development so it hits bumps from time to time that might effect someone who makes a living with their rig.
This is what I have a big problem with. What flight controller does not have issues? They ALL do. Even the $100,000 Aeryon Skyranger does. An operator told me you have to recalibrate the magnetometer every time you change the battery. Even if flying from the same location. I also witnessed how the control link drops out all the time.
There is absolutely no solid indication that Pixhawk is less reliable than anything else. I think it is the very fact that we DO admit to bugs, and fix them in a very visible way, that make some people think it's less reliable. If we had a policy of "never admit to bugs, ever" as others do, would that make us appear more reliable to the market? Probably. Unfortunately this rewards this behaviour, which is why some companies do it.
Yesterday I posted a great example. A developer found a problem with Arducopter that's been there for a long time. They have been quite busy so it's just now been fixed.
Which issue was that? The one I just mentioned about a baro glitch? That should NOT be an example of how the system is unreliable, or not ready. That bug has been out there... pretty much forever it seems. We've never seen it before, which would indicate that it's a once in a billion problem. The fact that we publicized it and fixed it does not mean the system is less reliable than one where the system designers would simply hide the whole thing, and roll out a fix in the next release without ever saying anything. You do not hear about once in a billion bugs from closed source systems.
But it makes the point that things just need a bit more time. Or as Robert mention, maybe some more funding for full time developers. I don't know if the devs will ever have time to lock the software down and come up with a highly stable production release.
This is very wrong.
Even a $100,000 UAV system can't release a "stable production release" and have it all work right perfectly. Your expectations are WAY too high.
You don't think that Boeing has updated the software on the Dreamliner a few times since it's release? Get serious.
It's when systems
stop updating firmware, that you know you have a problem. My Sony Xperia Z1, gets updated constantly. It's now running Android 4.4.4, fully up to date. Runs like a top. My Asus Transformer Infinity, hasn't been updated in a year and is stuck back on Android 4.2.1. It's a piece of crap, and barely runs. I'll never buy an Asus Android device again.
@R_Lefebvre - Were you involved in the hardware design of the Pixhawk?
No. These things tend to drop on our laps after they're pretty much done. Lately, they have actually asked us what features we'd like in the next hardware, but that's about it.
I've always wondered why they didn't go with a separate IMU unit with anti vibe mounting like DJI and XAircraft? I'm not an engineer but I've always wondered if having the sensors on the controller board might not be optimal? What do you think? I know that vibration has always been an issue for the APM/Pixhawk and it has to be mounted carefully to account for this.
My opinion on that is... it's hard to do properly. And I guess they didn't have the right person who could pull it off.
Also, what' up with OSD? That mini OSD from a third party vendor is just a pain IMHO. It would be nice if the Pixhawk had a more integrated solution for that. Do you know if there any plans for that? 3DR is usually doing some pretty interesting things in the background. I know devs can't breech NDAs but I'd be interested in your opinion about this.
I don't really see what the problem is with it?
I doubt you'll see an integrated solution. The problem is, the more things you integrate, the more the system becomes an "all things to all people" hardware. It starts to get really large, really fast. What else should they integrate into the board? A low voltage power supply? No, it should have a 12S rated power supply. Then people with a 250 quad, get stuck with an expensive 50V system. Build in the sonar? The Lidar? Built in ESC's? I think you're going to see more peripherals in the future, not more integrated stuff. A system which offers so much capability can't start building in all the hardware for all the users.
Eventually the ESC will be a peripheral, driven from UAVCAN communications just like all the other peripherals. Then we can do away with all the garbage PWM devices still in use. Fully digital, all the way. This has already been demonstrated.
I've been wondering who is the driving force behind Pixhawk. Is it 3dR, the owner of Wired magazine, a few independent engineers, or who?
The owner of Wired has nothing do with any of this. Chris Anderson used to be the editor or Wired, but quit over a year ago.
The Pixhawk design is a joint venture between 3DR, and the PX4 group at ETH, a university in Switzerland responsible for cutting edge UAV development (Rafaello D'Andrea, the guy with the crazy quad videos works there). ETH mostly does the conceptual design, and 3DR makes it a production reality. Sometimes they ask us for input or show us prototypes. We usually get beta hardware.
I think the answer to that would determine how far and how fast they can go because it would establish the funding levels. If 3dR is the prime then further expansion and refine won't come about until sales increased to a level that could support both the hobby side with their small MR's and R&D at the same time. Staffing is expensive and although they helped keep that down with the production facility in Mexico it's still a cash heavy operational requirement.
Yep. And the problem is that, well you guys don't see it here much, but there are tons of Chinese shops selling cloned hardware that undercuts 3DR's pricing. It's actually totally legal to do this as long as you follow the licensing rules. No royalties, but you can't infringe on 3DR's trademarks (the Pixhawk name, 3DR logo, etc.), you must disclose that your hardware is a derivative of an open source project, and if you distribute the code, you must also disclose that it is part of an open source project, and publish any changes you have made. There are several companies obeying the rules, but their pricing still hurts 3DR's ability to put money back into development. And these companies are not themselves contributing to development of the hardware or the software. They are leeches, IMO. This is not what open source is supposed to be about. But then you also have companies selling under the Pixhawk name, and some even put the 3DR logo on the hardware which makes it hard for people to know who they're really buying from. Then you have Walkera which is using modified versions of Arducopter on two of their RTF machines, without disclosing such.