First Impressions of the H520

PatR · Sep 10, 2017

The principle is quite feasible, and one used extensively with military grade UAV systems. It is also extremely expensive to incorporate and requires a massive level of computational power. A system pretty much requires a computer program, or separate computer, dedicated to target lock technologies. The camera/gimbal system also needs to be far more sophisticated, inclusive of an image adaptive smoothing program, than what small multirotors are capable of providing at a price point the vast majority can possibly afford. At super zoom levels the gimbal is not effective for stabilizing the image, it's done through image stabilization programs as the imagery is received at the ground station.

arruntus · Sep 10, 2017

PatR said:
The principle is quite feasible, and one used extensively with military grade UAV systems. It is also extremely expensive to incorporate and requires a massive level of computational power. A system pretty much requires a computer program, or separate computer, dedicated to target lock technologies. The camera/gimbal system also needs to be far more sophisticated, inclusive of an image adaptive smoothing program, than what small multirotors are capable of providing at a price point the vast majority can possibly afford. At super zoom levels the gimbal is not effective for stabilizing the image, it's done through image stabilization programs as the imagery is received at the ground station.

I don't agree with what you say Pat except to stabilize the image in real time for what it would need computational power. Let's forget for a second about stabilizing the image and just let the camera point to the target. In the video of the message the gimbal controller is a Strom32 that has a retail price of 20€, or less. Produced by a big company like Yuneec the price would be ridiculous. The gimbal controller communicates with a Pixhawk controller and do all the work. Even if the H520's Gymbal controller does not have that functionality, the H520 controller could do it. I think we already have a version of the Pixhawk controller inside the H520. Both controller and software are open and Yuneec has made her own version version which allows to have the announced functionalities.

Without knowing in depth the controller of the Yuneec Gymbal, I believe that it would be feasible as a programming job and then to assemble the external device and establish a link with the aircraft. From my ignorance I think it is feasible and cheap. At least I would like it and the versatility and possibilities it would give us would be unmatched.

Tuna · Sep 10, 2017

arruntus said:
Without knowing in depth the controller of the Yuneec Gymbal, I believe that it would be feasible as a programming job and then to assemble the external device and establish a link with the aircraft. From my ignorance I think it is feasible and cheap. At least I would like it and the versatility and possibilities it would give us would be unmatched.

The problem is one of accuracy, which comes down to the positioning information for the drone, the target and the gimbal attitude. If you watch videos like the one you posted, or using the Wizard, you'll quickly see that the targeting is really quite approximate. That doesn't matter when you have a wide angle lens as the subject stays in the frame. When you have a high zoom level, the target can very quickly go out of view, and even small motions make very large changes to your viewpoint. So to work with a high zoom, you need fast processing and smart targeting, which is really non-trivial.

You also need to consider the use that it'll be put to. If it's inspection or search and rescue, you can't put a GPS tracker on the thing you want to see. Inspection needs one type of tracking (hold view on the object being inspected) whereas search and rescue needs another (ensure all areas have been viewed), so there isn't one single solution for the job.

Of course, all of these things can be done, but it adds more expense and complexity to the system. If a camera has specialist application, it's going to have a specialist price.

PatR · Sep 10, 2017

After being immersed in the target lock/tracking high magnification field for the past decade the difference between theory and reality, especially where cost comes into play, is frightening.

arruntus · Sep 10, 2017

I agree with both of you.

Yes, with zoom would help but it would still be difficult to work because the movement of the aircraft is magnified, there is no doubt but any help always comes in handy. Gymbals would have to evolve much more to make it possible.

Yeah, Pat, if I agree with you, but I was just talking about the camera following the target, without zooming or stabilization or anything. I would also like to do what you say, but in this segment of prices today... I think that for a few years it is beyond the possibilities. Everything will come.

We're going to ask little by little

PatR · Sep 10, 2017

One method of locking on a non moving target location, a system will drop a GPS designator at the center of a target box slaved to the lens focal center. The focal length can change but the target box remains fixed to focal center. The longer a lock is held on the location the better the GPS resolution becomes. The system references the target location GPS tag to the aircraft's moving GPS position. GPS resolution improves over time when certain things are done but I can't share any of that. For a moving target "hard edges" come into play for a target lock but they don't function well until the target area is sized correctly. The target area is not the lens' field of view but an adjustable target area that resides within the field of view in use at the moment. As the camera is zoomed in small vibrations that weren't visible farther out become a real problem. At 20+ zoom the vibration can render the imagery unusable. There is also the "digital zoom" factor to deal with. At 70-100 power it takes a computer make sense of the image from a reasonably well designed UAV and gimbal system. It's pretty complicated stuff, but as you say, the technology in this area moves awful fast. Perhaps someone will come up with a gimbal that self corrects balance as the length of the lens changes.

arruntus · Sep 11, 2017

This is the device the brothers created.

Basically it is a GPS receiver, a transmitter that transmits the position to the aircraft and a battery.

The drone receives the position of the device or target, calculates its position and sends the command to the gymbal to aim. If we look at the video at all times points to the cyclist who wears the "clock". The target is always in the center of the image. The gymbal does two functions, its own which is to stabilize the image and receives turn and tilt commands.

Any gimbal has these two functions, we can move it manually, instead, the controller moves it automatically based on the position he gets from the "clock" gps. This project was made possible thanks to the Gymbal controller because it was able to communicate with the flight controller. The other gimbals didn't have that functionality.

They even created a program to control the gymbal from the cell phone. This project unfortunately was parked more than a year ago but if two brothers in his house have been able to do it.... a company like Yuneec will not be able? Another thing is that Yuneec want to do it....

The idea is launched, time will tell us if it comes to fruition.