I first have to express that most any strategy can “work”.
This is how I concluded video to be the best for OKE at the moment.
Close your eyes and imagine flying a multi line power kite, as compared to a “kite drone” attached with a single tether. Which would be harder to fly? Your senses don’t have enough data to fly the drone (well), you would need some type of flight controller to take care of certain variables. Whereas kiteboarders (with multi line kites) are taught to not stare at the kite.
This is to say I think it is possible for a multi line kite to be controlled, just like a competent human with tension control. You can be reasonably certain if the lines are of equal tension with a sturdy kite, that the kite will at least not crash (in most conditions.)
Now to pair this with visual sensing, we would train a neural network to predict the possibility of which pixel the kite will show up on next. A foil will behave in a similar way given the load and tension on the lines. If the tension is X, you can assume the amount of lift the kite produces. If the kite is moving across the pixels at a certain rate, (given the length of the lines stays constant) you can determine the speed and lift of the kite.
This method only works for multi tether craft because they naturally fly on the surface of a quarter sphere, wind-window.
As for the camera being expensive. The kite engine is mounted on a turret that always faces the kite (passively), on which there are rails to attach a camera. Using a fisheye lense, and a constant length of line. Many things can be determined visually and even more combined with tension sensing.
Automated control would be very much like teaching a person to fly a kite, albeit through video.
This example is ten years old now
"There is a function in openCV called “line” which will draw a line between two specified points. The angle of the line can then be found easily by trigonometry. "
Looks pretty clever. Obviously, a lot of thought and work is going to this. Sometimes one dedicated person can outperform large teams. Would be nice to see it actually pulling a boat. Especially faster than existing high speed sailing configurations. Maybe it already has, in another video - don’t remember if I’ve seen it.
Yes, generally speed records occur crosswind, the kite moving with the boat. It can be the same for a “classic” boat. However the same kite flying by crosswind figures (rotating or by figure-eight) when the boat is going downwind produces more power (as for any electricity AWES in yo-yo mode), but this is not good for speed records with the same special floats, although correct with a classic boat, the kite rotating acting like a spinnaker. So, as you indicate (if I remember correctly) OKE device with only one kite can also handle all gaits crosswind or downwind.
Crosswind both float and kite (1:34):
Crosswind kite but in probably downwind gait for a catamaran (1:12):
There are some really great conversations going on on the YouTube. What do you think of this:
“…Sailing tangent or “Cross wind” or Beam Reach, we would keep the kite relatively static/steady. Correct. The looping would be the most beneficial for downwind runs or producing electrical power (preferably at anchor). With going down wind, consider this… (and I’m not sure, which is why I am loving your questions) Instead of a typical jibe in a zigzag directly down wind… think of a zigzag at more of a 45-degree angle. One that osculates with the looping motion of the kite… The kite line will have to be pulled in if there is slack, and let out if overpowered… It will indeed be a delicate balance.”
This possibly demonstrates a combination of the two flight schemes mentioned above
I think I could agree with what is written until “zigzag at more of a 45-degree angle” (I don’t know).
A simple description of the three modes of kite work, includes a sketch in page 5, and summarizes my previous comment in short:
We will focus on modes 1 and 3.
Mode 1, described as static crosswind kite flight, boat perpendicular to the wind: medium traction.
Mode 3, described as dynamic, crosswind kite flying in figure-eight, downwind boat: strong traction.
For speed records the mode 1 is used (first video). For strong traction or electricity generation AWES, the mode 3 is used (second video). As you know, we can replace (I think advantageously) figure-eight with rotating figure.
In this way it would be possible to implement loop-figure in yo-yo mode.
That said some adaptations would be made, to allow the free rotation of the OKE device: this should have a central hole allowing it to slide through the bar (then central) of the KGM thus modified, while it rotates around said bar. The bar can be tilted.
During the reel-out production phase (several loops) the OKE device moves forward, then during the reel-in depower phase the OKE device moves backwards while it stops rotate.
This idea of looping pumping a linear generator likely suffers from a lot less line wear/fatigue than a system running the lines over sheaves (see Rishikesh’s talk)
Here are a few observations and a question which I start with:
Can it (is it reasonable to consider) automatic launching the kite? I mean ok, there are advantages of not having any electric device up, but most rigid wing single line designs do implement automatic launching.
The two observations:
Regarding the disadvantages of radio/sensors on the kite (or a pod under it) - I have a tiny drone which the flight controller itself weighs a few grams. The whole drone flying weight is 33 grams, battery is about 12 grams and the electronic pcb board maybe 3 grams, which includes elrs radio capable of several km or range bidirectional, 4 brushless motor drives with a total power of >20watts and an up to 400mWatt video transmitter. I mean that simple sensing (orientation & altitude) and LEDs (for night vision) would need very little power and be also very reliable. There are similar radios with dual band capability, or one can add a second radio for one more gram penalty. These are used in very expensive (and dangerous to fail) professional heavy drones.
Seeing kite’s position with a camera is almost a non issue, the algorithms can be very simple since the kite is pretty much the only thing in camera’s FOV, it is not like it is hidden in a forest. But there can be some tricky situations like seagulls playing around or, more seriously, the sun shining right behind it. Also having it painted with a contrasting, asymmetric unique pattern would help with estimating its orientation if kite-attached electronics is such a dreadful perspective. During night, instead of trying to illuminate it with a flashlight or lase I would rather recommend a pair of LEDs one at each tip of the wing or at the two control line knots where the bridles fan out. That would allow the visual algorithm to estimate both orientation and distance to the kite. Fog I doubt to be an issue since when it happens the wind’s speed tends to be missing too.
I think controlling from the ground might be easier for kiteboats:
you can not use loops for most conditions so you need to fly eights (or static), which requires more control power and would request either an heavy pod and power feed through the line (or onboard windturbine) or to use very long lines to gain time for reacting.
waterproofing might be needed in case of “landing” in the water.
sensor weight was an issue at small scale, but video processing speed/latency was as well. But we gain an order of magnitude on both in the last 10 years, so it will likely not be an issue in the next decade. We will probably have both for redundancy.
Sensors alone high on the kite are not that heavy and are reliable.
The reliability / weight become significant with flying actuators - which are required in case of single line kites.
An advantage of boats (vs ground base stations) is a boat can move, steer and maneuver so robotic launching operations should be more easy to implement. Even if only for the reason that a boat can add apparent wind speed by moving upwind till the kite gains sufficient altitude.
