Flight modes on a tether

[moonman]

I'm working on a tether system now for personal use, as the only ones I've seen so far are in the $30k-$100k range. I am past the point of the power supply at this point, and just trying to figure out in my head how this is going to work , if it does end up working at all lol. Here's what I'm trying to figure out.

So you've got a tether running to the middle of the bottom of the drone. One way to handle the wire is to have it always have slack in it, and wind it in as you're landing so as to keep it out of the way. BUT that's not how I want to do it. I want the drone to fly straight up in the air and just hang out there... when it's time to land it, I'd like to use the wire spool to actually winch it in, and have it land on a platform above the spool.

I'm using PX4 (that's the plan anyway) and looking at the flight modes. It's almost like i need a passive type of altitude hold mode, always pulling up, but not so much that the motors are at full bore the whole time. This way it flies to as high as I let out the cable, and allows me to crank it in. If this makes any sense to anyone lol.

Any insight would be much appreciated.

 

[moonman]

Here's one I've been looking at for inspiration. You can see what I'm talking about here. It winches itself right back onto the landing pad, without using IR or one of those big printed things.

 

[wafflejock]

I don't see why this wouldn't work if just locked into angle mode with no input and a bit above hover throttle then the winch on the ground can just use a current limited mode (or velocity limited) to apply consistent down force to pull the craft to the platform with the tether under tension. It is a bit of a waste of energy to have a winch fighting the motors to some degree but not a big loss and simpler than using sensors or IR/QR codes etc. for localization.

 

[moonman]

I don't see why this wouldn't work if just locked into angle mode with no input and a bit above hover throttle then the winch on the ground can just use a current limited mode (or velocity limited) to apply consistent down force to pull the craft to the platform with the tether under tension. It is a bit of a waste of energy to have a winch fighting the motors to some degree but not a big loss and simpler than using sensors or IR/QR codes etc. for localization.

Okay, let me make it a bit more complicated.. Let's say I want the platform to be able to move while the drone is flying. I still want the drone to stay straight above the platform as much as possible, but move with the ground vehicle. Also turn with the ground vehicle so they are both facing the same direction at all times.

And thank you for the response.

 

[MJW]

Just curious, what's the intended use of the drone?
Like for a follow me with an extended flight time?

And why not use fishing line, much lighter and heck why not a simple fishing reel?
Could set the drag on the reel very light,
Could put a mile of 4lp test fishing line on a saltwater reel.
But understand with an extended tether of any kind there's going to be motion drag of wind resistance as the tether moves sideways,

Does your drone have GPS?
That keeps mine right where I leave it, increase the mAh of your pack for extended time airborne,,,,,,,,,,,,,,,,,,,

 

[moonman]

Intended use is a bird's eye view on a boat. Maybe 100-150 feet up. my original plan was to use a GPS on the drone and on the boat, and do a GPS follow that way. Just match the one on the drone to the one on the boat. Thing is, I want it to be hands-off... The idea is to have a camera view that's facing forward and down at all times.

If you've ever used a drone on a boat, you'll understand the issues I'm dealing with.. landing is probably the biggest one, but the other is flight time for sure. If we hook into a large fish and the drone runs low on battery, it's an issue just dealing with the chaos of trying to do both things at once.

 

[wafflejock]

It feels to me like GPS follow is still the way to go most likely for it being tugged along at speed otherwise the quad will be constantly fighting to stay where it is as it's being dragged along and won't make for smooth video anyway. I think the tether and winch could still be done for power and landing purposes but think keeping slack in the line and doing GPS follow or otherwise using other sensors like optical flow or maybe Bluetooth beacons or other local localization transponders to keep the quad in range of the boat and to keep the tether loose.

 

[wafflejock]

If taught the tether itself would always point towards the boat but the quad orientation couldn't be guaranteed with this alone you need at least compass on both to keep them oriented in the same direction the tug direction could be used as a control input if the compass direction is known that and the tugging would be enough without GPS, in any case will be a lot of custom control hardware and software.

 

[gregb49]

Take care. When I was first fiddling with my JJRC X11, with it running, I didn't realise the implications of it fighting against me, until I let it go. This was the result:

Brick walls fight back!

 

[MJW]

Greg, This won't take very long now did it,,,,,,,,,,,

Kinda sad when the ad runs longer than the vid,,,,,,

Seriously, a PX4, assuming you're talking about a Phantom 4, it will max out on payload weight with just the power lead only a short distance up, wire weighs a lot when you think about the gauge of wire needed to drive it, we're talking pounds, not ounces, let alone gram's.
And they rate the lift capabilities in gram's for these drones.

When out fishing it's easy to think of thing's you could be doing because there's a lot of waiting time, and a cockpit can get very busy with a hook up so in reality why not have the drone at the ready on board then when a fish is hooked take off and start the "follow me" selfi mode, or just do the recording the event yourself as a pilot an let someone else deal with the fish, then when the fish is in the boat and things are settled land the drone, or net it as it get's close to the cockpit.
Now comes the tricky part, if you're the one catching the fish someone else need's to be a pilot, wind over water is another deterrent against a tether,

Larger mAh battery would be easier for the drone to lift than a tether & power wire feed.

Just trying to be realistic, sorry.

 

[wafflejock]

I believe they meant a pixhawk 4 flight controller since talking about flight modes but could be wrong... My new hexacopter one prop and motor lifts about 2lbs at full throttle so full up force would be something like 12lbs but still mostly agree this sounds more of a challenge than it's worth... Net it as suggested or build a big enough fold out platform to land on without damaging the boat or ditching the quad in the sea (and get some floaties just in case).

 

[GJH105775]

If you're up for alternative ideas, a large rod with what is basically a single-axis camera slider might be a more efficient way to do this, the issue, of course, is keeping the video stable. You could use a PID controller, or even a quadcopter controller to counteract the up and down motion of the boat, but keeping any reasonably sized rod from fluttering in the wind is not an easy task.

You could also make it a 3 axis to make the video more stable, again still using a simple quadcopter's flight controller (or any board that would support a 3 PID controllers, and 3 motor drivers). At this point, it is basically a 2 axis GoPro/Runcam gimbal on a 1 axis slider.

Here is a quick image that I put together to help explain this. The flight controller/gimbal controller and camera would sit on the blue arm. You probably would need to use small stepper motors as apposed to servos for something like this. If you weren't making it a more permanent fix, then this would not be an idea worth the effort.


I've seen people use DJI sparks on boats, but it's something you'd want to toss up once you knew you needed it, and landing could get hairy. They have a pretty good follow mode, so you'd get much better video than just a vertical camera slider.

 

[moonman]

Greg, This won't take very long now did it,,,,,,,,,,,

Kinda sad when the ad runs longer than the vid,,,,,,

Seriously, a PX4, assuming you're talking about a Phantom 4, it will max out on payload weight with just the power lead only a short distance up, wire weighs a lot when you think about the gauge of wire needed to drive it, we're talking pounds, not ounces, let alone gram's.
And they rate the lift capabilities in gram's for these drones.

When out fishing it's easy to think of thing's you could be doing because there's a lot of waiting time, and a cockpit can get very busy with a hook up so in reality why not have the drone at the ready on board then when a fish is hooked take off and start the "follow me" selfi mode, or just do the recording the event yourself as a pilot an let someone else deal with the fish, then when the fish is in the boat and things are settled land the drone, or net it as it get's close to the cockpit.
Now comes the tricky part, if you're the one catching the fish someone else need's to be a pilot, wind over water is another deterrent against a tether,

Larger mAh battery would be easier for the drone to lift than a tether & power wire feed.

Just trying to be realistic, sorry.

I left out the part where the point of the drone is to help spot the actual fish swimming on the surface. I don't care about filming the fish fighting... Part of the point of the tethering is so that I can just leave it up there when fighting a fish.

 

[moonman]

Maybe I need to go this route with image detection and/or IR lock. These guys are able to pull it off from moving vehicles and boats... but they are targeting the military and a bit outside my price range.

 

[wafflejock]

I know there are/were openCV libraries for quickly detecting QR codes and determining their angle etc to do augmented reality (AR) type stuff. I used it to try and do some 3D scanning using a Kinect for the point cloud and servo to have the program control moving the thing being scanned, it sort of worked but point cloud from Kinect v1 was rough.

I think ideally nowadays would use binocular cameras (two spaced a known distance from each other) and can either triangulate the distance once the angle of the QR code from each camera is known (and rough distance), or else go the full modern route and use machine learning but that requires tons of data and or high control of the sensor input/environment or filtering to make it all seem the same on the input side of a neural network (for training and inference to match). In any case yah this is a serious undertaking.... Using just a known IR blaster array on the ground should work too for the "last mile" to the landing point and would be simpler (it worked for Nintendo and Sega with their guns don't see why it wouldn't work as well today).

---

If programming inclined and/or interested in the Neural Networks solution check out "darknet" from pj reddie, it is an open source nn implementation in C that can utilize CUDA cores of GPU for reasonable speed training and allows for customization of the network but is pretty easy to get started with training it using image or text input and preexisting network configurations.

 

[MJW]

After watching the linked vid it seems you be better of with an tether free drone so it could get out a ways from the boat in all directions, the camera's field of view is pretty limited, a camera drone can go out in a circular pattern and cover a much greater area in a shorter time, but most likely spotting fish from the air is going to be a bit difficult as that hey blend in real well with the shade of deep water,,,,,,