Producing Energy with Drones - Wind Drones


One of the most interesting uses for drones is often overlooked: producing energy.

Yes, drones can produce energy. So much so that companies like Google, Shell and E.ON bet on drones to become major producers of energy in the not-too-distant future.

How do drones produce energy?

A fully autonomous plane-like drone (like the one pictured above) is attached to a tether and is flown in the wind like a kite:


The drone produces wind energy either with mini-wind turbines mounted on the drone (first picture) or simply by pulling on the tether and turning a generator on the ground (pictures two and three). More details here.

The advantages of such Wind Drones over wind turbines? They can be built with only 10% of the material needed for a wind turbine. More importantly, Wind Drones can reach the much stronger and steadier high-altitude winds that are literally out of reach of wind turbines.

Wind Drones bring the digital revolution to the energy market, the largest market of the world.

The result could not only be a potential "magic solution" (Bill Gates) for our energy and climate problem. The result could be the commercially most important drone application, the "Trillion Dollar Drone".

Most companies in the field of Airborne Wind Energy build rather large prototypes with industry grade components. We at Daidalos Capital were in addition interested in rapid and cheap prototyping and supported University of Bonn in building a mini Wind Drone prototype out of a standard model plane, a Pixhawk and modifed ArduPlane software. 

The project was a success and we managed to prove controlled tethered flight with a prototype costing less than USD 1.000 in material costs.


More details on the ArduPlane project will be available soon. 

Learn more about Wind Drones and their advantages over wind turbines here: “Producing Energy with … Drones?!”.


E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones


  • @Greg: do you think about kites that are attached to the ground via tether(s) or freely flying ones? In the former case, some airborne wind energy companies/groups were using them. They have much lower lift coefficients and are easier affected by UV radiation than rigid wings, though...

  • I am now wondering if ArduPilot can control a parafoil? 

  • @Andrew, thank you!

    The paths are given as intersections of the sphere with radius=tether length and planes, i.e. the direction perpendicular to the plane and a distance from the origin=home position. The great and small circles are obtained form planes with zero and non-zero distances from the origin, respectively. The figure-eight is built in the ned system with the geodesics crossing point directed in negative z-direction from home. Given demanded arc-lengths of the geodesic segments and an angle at which they cross, we calculate the small circles by demanding that the unit-tangent vector is smooth (the geodesic curvature isn't, though).  This whole system is then rotated in wind direction and inclined by a desired angle. So far, we determine the wind direction prior to flying the eight from the airspeed-calibration data.

    Implementing an automatic adaption to changing wind directions is not yet our highest priority, since we have many more things to work on. But it is of course essential for a complete system. Also, an automatic flight-path optimization that maximizes the power output would be nice. One of the urgent things to be done is to clean and modularize and document the flight modes so that they are more readable. We are also working on a paper that explains in more detail what we have done. The figure Udo has posted is extracted from the AHR2 and ARSP lines of the log. We will add more analysis in the paper.

    Very cool that you mention the SITL: before the first flight we have simulated the figure-eight pattern, also using a simple model for the tether force (we added it as external reaction to the plane model). We have not added such a model to the airplane-code yet and were amazed to see that the controllers are still able to keep the plane stable -- albeit of the extra force. Adding a tether model to the Plane code is also something we think is of high priority.

    For takeoff, we use loiter-to-alt with a radius that allows us to smoothly dive into the figure-eight pattern. This switching of flight-modes still is manual. Also, landing  is manual. We have to get some experience of how to reel in the tether before also automatizing this... another construction site.

  • Developer

    @Christoph, interesting!

    How do you parametrise the flight path? Do you set a radius and mid-point? I assume it has to automatically change position with wind changes, and possible change radius of the circles with wind speed?

    I'd love to see the modifications to ArduPilot, and if possible a dataflash flight log. It would be interesting (and probably fairly easy?) to add a tether like this in SITL so we could fly a wind power drone in the simulator.

    How do you handle takeoff and landing?

  • @Andrew:

    Thank you very much for your interest. I worked on the technical side of the project.

    We wanted to fly a fixed-wing aircraft on a tether of constant length in a figure-eight pattern that is inclined at a certain angle.

    For this, we mainly had to:
    - implement a mode for flying on an inclined circle (segment) (a type of loiter mode with locally variable altitude)
    - implement the figure-eight path on a semi-sphere, constructed from two great circles (middle parts of the        figure-eight that cross each other) and two small circle segments (downward loops at the right and left end of each figure-eight) and switching between segments.

    Moreover, we had to modify and reinforce the wings and fuselage of the
    aircraft (Easy Star II) with carbon fiber (the goal is to pull on the tether
    with a large force much higher than the nominal payload). It is shown below.


  • Developer

    nice! I'm looking forward to hearing more about what you needed to change in ArduPilot for this application

This reply was deleted.