Back to the homepage - Hans-Georg Michna

SailBot—a robotic sailboat

Last change: 2012-12-15. Copyright © 2004-2023 Hans-Georg Michna.


ON'T ask me how I got the idea, but I decided to try to invent a GPS-navigating robotic sailboat. The rough theory is finished. I'm thinking of a mostly submerged vessel with a ballasted keel and only one sail consisting of an airfoil, looking a bit like an upwards protruding paddle, that can be moved in two axes, rotation and forward-backward inclination. No rudder, but a GPS antenna. Control is similar to a windsurfing board.

The SailBot

The aim is for simplicity and ruggedness. Such a vessel could transport some load over very long distances at a leisurely pace and actually quite quickly in strong winds with a favorable direction. It could also serve as a fixed buoy, sailing a figure of 8 all the time to essentially keep its GPS-defined position.

If the robotic vessel is too sluggish to tack, it would be programmed to always jibe instead. No big deal, as it could then sail mile-long tacks on the open ocean.

Such a vessel could transport some load over very long distances

Arriving at its destination, it would again sail a figure-of-8 pattern until collected.

It could be designed to be stable in relation to the wind direction to minimize energy consumption, i.e. it would adjust its direction, then freeze and sail roughly straight for a long distance without using any electricity at all, while accepting some deviation from the programmed track. The next adjustment or tack would then compensate for the accumulated track error. This way the vessel could move on very little energy, possibly economically enough to be recharged from solar panels during the day. But this is probably not required, as controlling the sail wouldn't use much energy anyway.

The navigating algorithm should work like this. The vessel is preprogrammed with a starting point, a destination point, an ideal course, and an area around the ideal course, possibly a very wide area, inside which it is allowed to move.

A wide area has the advantage that it doesn't have to tack often when going against the wind. If possible, it will go in a straight line, of course. The algorithm could be this. Make a new decision, trying to sail straight to the next waypoint along the ideal course. If the wind doesn't allow that, determine the nearest angle the wind allows while optimizing speed made good and keep going straight. If the wind changes significantly, recalculate and make a new decision as if starting from the current position. When reaching the border of the allowed area, tack and make a new decision again. This is relatively simple.

A bit more complex are tack and jibe operations, but I think a clever self-learning algorithm could be devised, which would automatically adapt to the actual conditions, payload, weather. This needs some experimenting. The vessel could always tack and accept being driven backwards under unfavorable conditions while keeping to turn. Or it could always jibe, which would be much simpler, but possibly less efficient. Some experiments will quickly determine the best way to do this.

There could be a separate bad weather algorithm, such that it would merely try to stay near the ideal course with a minimum of control energy to wait out bad weather. The algorithm could, for example, make sure that it doesn't go downwind at high speed in a storm (unless that direction is desired, of course), for example by assuming a sail position slightly aft and at a nearly right angle to the main body axis.

Questions by Mark Borgerson and my answers

Question: If your vessel is semi-submerged, how will it react to waves where the wave height is a significant fraction of the sail height? It would seem that you would need a lot of steering engergy to overcome turbulent wave action at the surface. If you have ever tried to steer a normal sailboat down wind with the seas behind you, you know what I mean about steering energy.

Answer: I don't know. I'm actually thinking of a sail that is much smaller than wave height in bad weather. A very simple assumption is that the vessel tries its best to maintain course. If it really can't make it, it can try to let itself be driven back along its course, rather than wandering about aimlessly. Essentially it would wait for smoother weather.

I could imagine that, as long as it can roughly keep its body's orientation, it may be able to keep going even in rough seas if the wind direction is not too adverse. It's got to be rugged enough not to break even in the worst weather, of course. Another possibility is to make it almost as dense as water using ballast or more payload, so it would always dive under big waves, so it is not thrown about so much by wave crests.

Question: I think there are critical scale factors relating vessel length, wave height and period, and sail height.

Answer: I'm envisioning a sail that is relatively small in relation to the overall size of the vessel.

Question: How is this robotic vessel going to obey the rules of the road? A 5-m long research glider which spends most of its time far below the surface is a lot less dangerous to passing surface traffic than a 50-m multi-ton vessel with no maneuverability and no clue about surface traffic in its vicinity.

Answer: My idea is a pretty small one, no longer than 1 m, possibly even shorter, only the size of a model boat. Too small to make a dent into other ocean-going vessels. And since it would normally sail the high seas, it wouldn't be a danger to small coastal boats or to anything else. It would be comparable to a piece of wood floating around.

Comment by Mark: I think you are going to have severe control problems. The first 2 meters of water and the first 2 meters of air in the open ocean are very turbulent if there is any wind above a few knots. Any breaking waves are going to toss your boat around severely. I would expect that it would be able to sail a desired course only in winds less than 5 knots and seas less than two feet. Those conditions are pretty rare in the open ocean. And with winds that low, there will be very little energy to extract in the first few meters above the water.

This turbulent boundary is the reason that I earlier said that there are problems of scale to overcome. A 100m craft can have a taller sail and deeper keel and would be controllable in a much larger range of sea states.

Question by A.: It is probably less complex to keep the sail about a single contollable axis and a seperate rudder for yaw, at least for a first venture.

Answer: When you consider that you need some kind of waterproof gasket for each moving rod that comes out of the fuselage, it may be simpler to have just one.

Mechanically a two-axis control is not all that complex.

But there is yet another reason. The rudder is ineffective while the vessel does not move. For example, when the vessel tries to tack and can't turn around quickly enough, it begins to glide backwards. This would require the rudder to go to the other side at the right moment, which adds some complexity.

With a single sail, this does not matter. The sail goes forward and veers to the other side at roughly the right time during the tacking maneuver, and the thing will tack to the other side. It may glide backwards under unfavorable conditions, but it will keep turning and eventually go in the right direction without much further movement of the sail. The only remaining movement is that, when the vessel has turned far enough, the sail has to move back aft. (The neutral position of the sail while tacking against the wind should be somewhat aft, so the vessel keeps a stable direction without having to move the sail.)

You see, a few thoughts already went into this. (:-)

If you plan to actually build it, please hire me

Imagine the simplicity of using such a vessel as a buoy, for example a weather and ocean data gathering buoy. You could throw it into any ocean just anywhere on this planet, and a short time later it would reach its predestined place and stay there. With some radio communications you could even order it back for maintenance, without having to pay for a manned, ocean-going vessel to collect it and then to put it back in place. I'm pretty sure there are some interesting uses for such a robotic vessel.

Apart from the buoy task the only thing I could think of that it could be used for is smuggling (though small robotic planes would probably more suitable for that), but I'm sure there are some more useful and at the same time legal applications for it, not to mention military ones. As the thing isn't all that difficult to conceive, I'm sure that somebody will actually build it soon.

If you plan to actually build it, please hire me. I would like to help with the mechanical design and program the central computer for the SailBot.

Hans-Georg Michna


2006-04-21 – Popular Science reports on a somewhat similar device with a different propulsion system.

2009-02-13 – Gizmodo reports in its Crime section on The Rise of the Drug-Running Remote-Controlled Semi-Submarines. Those are not sailboats, but they confirm my earlier suspicion that remote-controlled vessels can be used for drug running or worse.

2012-03-14 – The company Liquid Robotics has invented the PacX Wave Glider. It resembles my invention, but uses wave instead of wind energy. Here is another article about it, reporting a transcontinental crossing.

Back to the homepage - Hans-Georg Michna


hits since 2007-11-01
Free PHP scripts by