A Remote-controlled, Autonomous Kite Generates Power

kite

Generating power from wind is easy – just stick a windmill on a pole and attach a generator. That’s not particularly cool, though, so [Adrien] and his team from his senior design project are using an autonomously controlled kite to generate power

The basic idea of generating electricity from a kite is to fly it around in figure-eights while unwinding the kite line from a spool. The very strong forces on the kite lines can be used to drive a generator which provides power for reeling the kite back in at a lower angle of attack. You can check out [Adrien]’s kite power theory page for a few more details on how this works.

Right now, [Adrien] and his team have a basic rig set up to generate power and are flying the kite via a joystick. Updates are coming, and you can check out the video of their RC kite in action after the break.

[youtube=http://www.youtube.com/watch?v=wXHFUfO7sG8&W=580]

33 thoughts on “A Remote-controlled, Autonomous Kite Generates Power

  1. Autonomous: “not subject to control from outside; independent.”
    Remote Control: “control of a system or activity by a person at a different place”

    Please don’t write “remote-controlled, autonomous” as a description for anything. They pretty much mean opposite things! This is one of my pet-peeves mostly because autonomous robotic projects are really hard, and it’s kind of frustrating when the label gets slapped on projects which pretty clearly aren’t (yet). Especially because it confuses people who don’t know the difference between the terms when they come to see a project on display.

    As a quick extra note, if you’re trying to say that the system is remote-controlled, but has a computerized component would be “fly-by-wire”.

    Terminology jerk: AWAAAAAAAY!

      1. Totally got that! And I wasn’t trying to suggest that it wasn’t (I even put that parenthetical “yet!”), but that still doesn’t mean that it’s a “remote-controlled, autonomous kite”. It means that it’s a remote-controlled kite which will eventually be autonomous.

        Alternatively: it is an autonomous kite which is shown here with the autonomous functionality turned off.

        Just don’t mash “remote controlled” and “autonomous” into the same sentence describing the same object. I gave the definitions to point out how they kind of negate each other.

    1. Incorrect.
      remote control
      : control (as by radio signal) of operation from a point at some distance removed
      autonomous
      : capable of existing independently
      The two are not mutually exclusive.

      1. So I see where you’re coming from here, and I clarified a bit of my gripe in the comments below. That being said, I’m more referring to the confusion generated when even very technical environments seem to interchangeably use the terms “remote controlled” and “autonomous.” This could be considered a “autonomous system” once they have the camera feedback installed, and the “remote controlled” part would be “capable of being remote controlled” as a side-note to the general autonomous part.

        Your definition of “capable of existing independently” unfortunately has one glaring flaw: independent of what? As in it’s capable of existing after the creator stop touching it? That seems like a low bar and would seem to make the use of the term “autonomous” almost completely pointless. In the context of building robots and other complex robotic mechanisms, I posit that the term “autonomous” is “capable of functionality independent of an external controlling mechanism or person.” I.e. the “autonomy” is that the kite or robot is capable of functioning in a complex manner long after human input has stopped.

        Now, there remains a problem with this definition too: at what point do you qualify “complex” or “long after”? For example, if I made a giant industrial carpet knitting machine that could run for 8 months given enough thread, is it autonomous?

        In general, I like to resolve this as saying that an “autonomous robot” is any robot capable of making a decision based on unknown stimuli. So essentially, the input can be modeled, but if you can give a complex task to a robot (i.e. cross this room) and it can figure out what all the obstacles are and make decisions about how best to overcome those obstacles, then it is autonomous.

        And as another note, there’s another question of that independence I mentioned earlier: what if the controls system and the robot are capable of being autonomous, but on in conjunction with each other? Is this still autonomous? As mentioned below, these are now being touted as “autonomous systems” because they have multiple parts. I personally say that a specific robot can’t be called truly autonomous unless it functions independent of everything (at least for an appreciable length of time). It has to be able to make at least one or two decisions independent of an external control system to be an autonomous robot in my mind.

        The downside here is that my definition of autonomous is a pretty narrow definition. It means that technically, if I’m being a picky jerk (which I kind of am, but I’ve gone too far to turn back now!), this is more of a fancy controls project than an “autonomous kite.” Note: THERE IS NOTHING WRONG WITH THIS. I’m just espousing my personal opinion on the subject of the word “autonomous” and you may take it or leave it as you will.

    2. Terminology jerk strikes again…I do agree with what your saying. When the argument is whether to call something a robot or a remote controlled vehicle, the line is a little gray. But autonomous vs. remote controlled is pretty clear.

    1. So this one is kind of a sticking point for me too, mostly because I think it’s a much easier problem to solve. So to stick with my definitions: this would be an “autonomous system which controls a kite” as opposed an “autonomous kite”.

      An autonomous kite would imply that the kite itself was capable of some sort of decision making and could control itself. You can see in the video though, all of the intelligence in the system is in the base along with the control mechanism. And their project description seems to imply that they’ll be using camera feedback to identify the position of the kite, so this seems even less like the kite has much to do with it.

      The reason I have that as a sticking point is mostly because of UAVs or “unmanned aerial vehicles”. These generally have some on-board computers which allow them to have limited control and do simple tasks, but the vast majority of their work is done by beaming all of their sensor data back to a computer somewhere else who crunches the numbers and makes a decision, or who hands over the controls to a human. There’s nothing wrong with this as a concept, I just don’t think you can say something is an “autonomous vehicle” when you can’t operate it autonomously! It requires something large and outside the vehicle to operate, and thus it isn’t capable of significant autonomous action.

      That being said, most of the companies who sell these types of things to the military have actually started referring to them as “UAS” or “unmanned aerial systems” to point out that you’ll need more than the vehicle itself.

      1. In regards to the kite though, something manipulating the tether of the kite could be considered part of the whole. As long as it’s all bolted to the tether, it’s an autonomous kite. Without a tether a kite is little more than a piece of refuse caught by the wind.

        1. That’s fair to a degree and I was trying to make clear that my overall gripe with the term “autonomous vehicle” being applied to different vehicles (such as the UAVs), not this project in particular. Once they get their camera feedback controls working and have the whole thing working I won’t really care if they call it an “autonomous kite” or “a system for autonomously controlling a kite” I normally try not to enter into these debates because I’m hopelessly semantic and talk way too much.

          That being said, I think you could build a kite where you tied three strings into the ground itself and put the pieces which pull the control lines in the kite and a small computer (say a beaglebone) with an accelerometer all on the kite itself and that would be an “autonomous kite”. I was just commenting that it’s a bit odd that this system really isn’t dependent on the kite it’s attached to, and saying that the kite itself was autonomous was a bit off to me.

          tl;dr: once they make it autonomous I don’t really care if they say the kite bit is autonomous, it was just a bit of semantic B.S. that my brain decided to spew. Sorry if I offended.

  2. Kudo’s for a video without some annoying background music.

    But….. Please stuff a rag in the mouth of the dork saying “chill moves” and other juvenile things. I’m assuming by “Senior design project” you’re talking about Uni not Junior high.

    //to be fair, I probably sounds just as stupid to my thesis advisers way back when, but was lucky that it was pre-smartphone video, pre-youtube days so it’s not documented for all time for future research teams, grant selectors, employers, etc to view and review//

    1. Obviously one wouldn’t want to general slang while writing a technical paper, or any slang for that matter, but I have to believe only a foolish person would hold a recording of someone using such slang against them in a moment excitement against them. Foolish because they just might reject a person that could make money for whoever it is they go to work for. While it never caught on around here, but in the early ’70s the word cock was used in a manner like far out- cool- neat-slick would have been used as adjective slang in some places around the US. To judge the past my the present would also be foolish in this context. None of this is to say all slang would be appropriate if it depicts some as a bigot racist or as sexist, but chill move doesn’t come close to that.

  3. And as a totally serious comment independent of the whole terminology thing: is there a reason you guys picked camera feedback as opposed to something that took the angle of the kite string relative to the base as the controlling feedback? I only ask because camera feedback is generally hard, error-prone, and computationally expensive.

    Maybe just have a three-axis accelerometer on the kite to figure out it’s angle to the earth?

    It might also just be that you’re doing it as a “minimum difficulty” thing for the project, so I’ll butt out after that comment.

    Neat stuff!

      1. So I was actually arguing that I thought your approach was harder than it needed to be.

        Why not take a pair of these slide potentiometers: https://www.sparkfun.com/products/9119 and put them at right angles to each other:
        one slide pot on the side
        | _
        one slide pot on the bottom (or top, whichever should be fine)

        Then make it so that the string which controls the kite will drag the two pots; i.e. if the kite goes up, the pot on the side should be dragged up, if the kite travels left or right the pot at the bottom should dragged in the same direction. Place this pair of pots at a fixed distance from where the string leaves the base. Then you can turn that x-y coordinate into an angle, and you can figure out how to control the kite using that as the feedback.

        I would think that would be simple, easy to build, and significantly more reliable than camera feedback. And the biggest plus (in my mind anyway) would be that you wouldn’t waste a ton of energy which you’re harvesting on a computer continuously trying to process frames!

  4. Perhaps slightly off-topic, but I always thought it would be cool to hook up a small generator to some on-kite LEDs for interesting night flying at the beach or what have you.

    Still an interesting concept.

  5. to their credit , after browsing the site it\s clear they are on the way to having it run autonomously. they have a video of them tracking the kite using camera+opencv so they are definitely on their way there and i think they deserve the title. great project! just to mention that robotics company festo had/has a similar project.

  6. Having spent years flying power kites, I’d be very interested to see a machine capable of doing it for a long period of time without any human input. Sudden gusts or lack of wind can require some very fast thinking and manipulation of the kite. I’m not saying it can’t be done, I just want to see it first :)

    This is clearly not autonomous, very misleading title.

  7. Saul Griffiths TED presentation lead me to believe he is speaking of wind turbines mounted on a wing shaped kite. Adrien and team’s approach to generating electrical is different, and will be genius if they can pull it off. I do doubt that their approach could generate near the power Griffiths method. Not that it matters for us in the US. Americans as a whole will run into the wall, and when we do there will some who will ask why didn’t anyone tell us about that, and it will still be illegal to wring the necks of those who ask ;)

  8. Why reel the kite back in? You could have some kind of actuator on the kite that changes its shape to reduce lift so very little energy is needed to bring the kite down. For example on that parasail kite, you could have a string going between the wingtips with a motor in the middle able to draw the string in, turning the front profile of the kite into a horseshoe shape. As the tips get closer together lift will be reduced. Then just the slack would need to be reeled in instead of having to pull the kite down against the wind.

    1. It helps in maintaining the tautness of the string. Too much slack and the string would break from trying to pull it in for maintenance. Also, with the wind blowing, that string will vibrate causing all sorts of guidance errors that have to be accounted for to ensure smooth sailing (yea, yea pun intended but is actually something that is very important in such applications).

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.