Pixhawk Feedback?

[Old Man]

If the yaw was to have "flowed" instead of "turn" that would have been awesome. I agree, so much time is spent seeking just "the" light conditions to make the scene as sensitive as possible. You missed the landing on the picnic table For some reason I have an affinity for landing on them when they are present.

 

[R_Lefebvre]

Yeah, the copter flight is pretty smooth. But the camera movements need a bit of work. Part of the problem is that we aren't natively controlling the camera, but in my case running through Alexmos. I really wish we had our own controller. But in any case, we can improve what we have. Ultimately, I'd love to be able to precisely script the camera movements the same way we do the flight path. At this point, you can set discrete commands like "look here", but the movements are a bit coarse.

 

[JoeBob]

Would you mind posting a screen shot of what your Mission Plan looked like?

Were all the waypoints spline?
Waypoint radius?
Ascent and descent speeds?
Did you use ROI?

I'm presently wrasslin' with a throttle/power issue on my Pixhawk/DiscoPro:

http://ardupilot.com/forum/viewtopic.php?f=21&t=9545

but autonomous flights where I'm only controlling the camera pitch and correcting horizon is where I'm headed.

 

[dazzab]

I'd like to make the yawing a little smoother yet. And with the automatic camera pointing (not shown here), the pitch also needs some work to make it fluid. But this is pretty good right now I think.

I've been watching the spline feature with interest. I have to admit that I don't really get it as every pro video I've seen is much smoother. I thought spline was more about smoothing between waypoints rather than smoothing turns per say.

Personally, I really like the repeatability aspect of automated flight but when I mentioned this to a friend his reaction was very negative and of course, technically it's not legal right now anyway. I managed to get yawing quite smooth on a two axis gimbal just by adding expo to my yaw. But again, my friends talked me in to using three axis with follow me on the yaw axis and so far it's looking great.

 

[dazzab]

So here is an example of Arducopter being used for artistic aerial photography. This video was shot using 3DR copters and flying Arducopter, for the purpose of creating an announcement about Richard Branson investing in 3DR.

Just lovely, I really enjoyed that. Thanks.

 

[Old Man]

Personally, I really like the repeatability aspect of automated flight but when I mentioned this to a friend his reaction was very negative and of course, technically it's not legal right now anyway.

We should all be paying close attention to this feature and be ready to join the fight that is likely soon to occur that will attempt to prevent us from using it. Automated flight and the use of way points is one of the foundations of sUAS operations among the major aerospace players. Without it they too are dead in the water. With the exception of the small hand launched craft like the Raven none of the players seeking ownership of the sUAS market is totally hand flown. Far from it, most are flown through various forms of automation. Some are too unstable to be flown by hand at any time. Even things as small as Raven and Dragoneye have the ability to be flown autonomously using GPS based flight plans, so why should we be excluded? I can name several so called "advanced" sUAS systems that are actually inferior in capability to what some of our FCs can do in trained and capable hands. Making that condition even worse is those systems use autopilots 50-100 times larger than ours, costing 1000+ times more. Believe when I say they are no safer in open airspace than our higher technology systems are, and they fail for the same reasons. The sands of Iraq and the bedrock of Afghanistan have been littered with the remains of many hundreds of "commercial" grade UAV's.

 

[jdennings]

Personally, I really like the repeatability aspect of automated flight but when I mentioned this to a friend his reaction was very negative and of course, technically it's not legal right now anyway.

Huh? It's perfectly legal, as currently confirmed by a judge's ruling. The fact the FAA appealed does not change that.
There's only one person pretending otherwise, Jim Williams of the FAA, and even him and the FAA lawyers know it's just pretense.
Why this meme/myth of illegality is still alive baffles me. I suppose the fact that the FAA, as a federal government agency, carries some authoritative weight and feels intimidating helps explains part of it. But the FAA does not have the power to make laws, only congress does, and certainly does not have any power when it comes to enforcing inexistant laws.

As far as using autonomous mission with spline for film, I am convinced this will be greatly used in the future. Once we get centimeter scale precision with things like Lidar and Piksi-ish gps, it will be a boon for precise, sophisticated shots. We just may not be quite there yet.

 

[dazzab]

Huh? It's perfectly legal, as currently confirmed by a judge's ruling. The fact the FAA appealed does not change that.
There's only one person pretending otherwise, Jim Williams of the FAA, and even him and the FAA lawyers know it's just pretense.
Why this meme/myth of illegality is still alive baffles me. I suppose the fact that the FAA, as a federal government agency, carries some authoritative weight and feels intimidating helps explains part of it. But the FAA does not have the power to make laws, only congress does, and certainly does not have any power when it comes to enforcing inexistant laws.

Sorry, I should have mentioned I live in Australia. Autonomous flight is not legal here. Pilots must be in total control of the copter at all times and LOS is required. It is possible to get exemptions from CASA but it's a hassle. I'm actually a big fan of autonomous flying but until it's _much_ more reliable I really think it's not safe.

 

[jdennings]

Ah .... good point. I should've checked your location next to your name.

 

[jdennings]

Speaking of advances in aerial filming and automation, here's a neat first step: visual tracking for gimbal control. With arducopter now running on Linux, seems like it's only a matter of time before precise POI and "automated" framing can be made possible. The possibility and no doubt upcoming availability of face tracking is going to make many observers quite unconfortable though ...

 

[R_Lefebvre]

Sorry, I should have mentioned I live in Australia. Autonomous flight is not legal here. Pilots must be in total control of the copter at all times and LOS is required. It is possible to get exemptions from CASA but it's a hassle. I'm actually a big fan of autonomous flying but until it's _much_ more reliable I really think it's not safe.

Are you sure?

In Canada, the rules read as if fully automatic flight is not allowed. In fact, it also states that the required system response from a loss of control signal is for the system automatically end the flight. However, I talked to the lead UAV regulations guy at Transport Canada, who stated that the rules were written in 2008. This was before automated flying was common. He stated that at this point, fully automated flying is completely legal, so long as the pilot always has the option to take control. So for example, if you want to do mapping, you set up your lawn mower pattern and hit go. However, you always have the option to take back control.

Aren't you at the OBC right now? Isn't that automated flying? I think it's actually also beyond LOS?

I've been watching the spline feature with interest. I have to admit that I don't really get it as every pro video I've seen is much smoother. I thought spline was more about smoothing between waypoints rather than smoothing turns per say.

The intention is smoothing the whole flight. The turns all blend. With spline, there are no straight lines. Now, there might be room for improvement, this is the first release of this feature. Heck, it's not actually even released yet. But the concept is here. It no longer flies straight lines, bouncing off waypoints like a game of Pong in the sky.

Also, as to smoothness... remember, this is not a shot from a large Octocopter with a Zenmuse gimbal. That was a Tarot 650 frame. AUW of about 2kg, but with 15" props, flies for 25 minutes. And as you can see in the trees, pretty breezy. I was actually pretty happy with the stability. That was also an action cam, on a 2-axis home-made gimbal with Alexmos. It's not even completely balanced yet.

So how much of the lack of smoothness was due to physics instead of programming? This is a test quad, not a professional video machine.

But as I stated, the yawing definitely needs work.

 

[Quinton]

I was about to take the leap on a pixhawk but then discovered the input voltage is only up to 4s
Are there any plans for a 6S version in the near future?

I have read about the Attopliot hack for 6S I am not that savvy with changing electronics.

 

[R_Lefebvre]

I'm flying it now on helicopters using 6S and 8S. It can be done, but you do need to know a little about electricity. I just power the Pixhawk with a high-voltage rated BEC.

I keep hearing about 3DR making a high voltage power module, but I don't know when it will be released. I think some of the alternate suppliers like Hobby King might already have these out? Not sure.

 

[dazzab]

Are you sure?

No, I am not sure. As you pointed out with the Canadian regs the language is a bit 'grey' in nature. But the old fart purists insist that any type of autonomous flight is a no no. In reality it's probably safer to fly with the autonomous systems but so far nobody has come up with any proof of that so people seem to default to 'old school' thinking/systems IMHO.

Aren't you at the OBC right now? Isn't that automated flying? I think it's actually also beyond LOS?

The Outback Challenge is on 22 Sep. Yes it's autonomous and BLOS. From what I understand it was quite a challenge to get CASA approval and insurance as a result.

Bottom line, I think like in the US most commercial operators go about their business maintaining compliance the best they can in such a fuzzy and quickly changing regulatory environment. The multi rotor hobbyists just do whatever they want and end up in the news from time to time when they do something stupid. It's all quite messy for the time being.

 

[dazzab]

I was about to take the leap on a pixhawk but then discovered the input voltage is only up to 4s
Are there any plans for a 6S version in the near future?

I seem to recall that 3DR now offer a power module that handles 6S. If not, it's coming RSN. But there are certainly knock off power modules on eBay as I have a couple. They support up to 6S. Using the power modules is nice as you get voltage/current readings in the GUI but I've powered Pixhawks from a BEC and then used a voltage/current sensor from Sparkfun. I believe that's the solution documented in the Wiki. Regardless, the Pixhawk can certainly be used with 6S powered copters.

 

[R_Lefebvre]

Yes, you can just use any hobby BEC, set to 5V, and plugged into the servo rail. You won't get current and voltage sensing without doing a bit more work. But that's all you need at a minimum.

 

[Quinton]

I see you can get different ones now with built in BEC like the following..
http://www.ebay.co.uk/itm/APM-Power...rt-30V-for-APM-2-6-Pixhawk-XT60-/151355843956

I take it you would have to have 2 lipos plugged in parallel through the one connector.
Is there no way to get the voltage/current sensing from 2 batteries separately?
I would love to give the pixhawk a try, but don't want to rush in if there will be a better solution in the near future.

 

[R_Lefebvre]

It is now sort of possible to have voltage from two separate batteries, but it is not possible to have current from two.

 

[R_Lefebvre]

Latest testing yesterday with 3.2 rc9 on my helicopter. From what I'm aware of, this is above military grade performance. Nobody has a VTOL UAV that can perform like this. :cocksure:

 

[SamaraMedia]

That's pretty darn impressive!