ardupilot goes into the water Part 11

The Start of the 2010 swimming exercises.
All told about the hardware...
All told about the software...
What´s next?



10A @11.5V for a speed of 5.6 km/h is lousy. The same speed with 5.0A would be OK. This benchmark was set by the Firebrigade-ship, but the catamarane had a much better straight-line behaviour.
OK, let´s improve the catamarane...

In the long winter evenings i googled a lot about catamaranes. Somebody had posted a comparison between different hull shapes / cross-sections. The one with the triangular cross-section as i was using, proved to be the worst in terms of drag. But it was the best in terms of straight-line behaviour and ease of construction. The best in terms of drag was an elliptic shape. Now i knew, where i was.


When googling for plans of model-cats i found this site:


This plan looked best:


There are also dxf-files available. A colleage of mine is a specialist for 3D CAD design. I gave him the dxf-files and he made a 3D model of the hull out of the DXF-files. (Thanks Matthias)
Here are the drawings of the segments:

But how getting a hull from the plans with the least effort?

diy-store -> Styrofoam with 100mm thickness --> some cardboard --> 14 cross-sections of the hull printed on paper --> some double sided sticky tape --> a hotwire styro-cutting assembly and three evenings when my wife had to watch TV lonely.

One of the hull-segments is the solution to the "guess-what" picture of the last episode.

This are the middle segments, stacked up.



All 14 segments (each with a thickness of 100mm) were glued together with styro-glue. A little bit sanding and the hull was nearly finished.

The glued segments, before the coating...

To add stability, the hulls were coated with a thin layer of fiberglass and epoxy.

The coating - and sticky fingers...

To connect the two hulls with a styrofoam-board that serves as a platform for the superstructures i used styrofoam screw-anchors and M5 Screws, that were epoxied and glued into the top of the hulls prior to coating.

The styro screw-anchor

To give it a real good finish there had to be much more sanding, priming and painting, but as already mentioned earlier, patience is not one of my strenghts, and so i decided to bring that provisional assembly into the water as fast as possible.

This construction is far from being perfect, but to prove the concept it was OK.

The first launch...

The first launch:
Yipee! I have invented the catamarane-oscillator!
The cat oscillated, but found its waypoints. On the first run...




The second one with decreased P ended up in my first swimming exercise in this year. It was still oscillating, but it did not manage to find anything. The problem seemed to be the lateral stability of the ship´s aft. The new hull had problems with his butt! Back at home i added fins on the backside of the hulls. The size had a triangular shape and went about 100mm below the waterline.

Back to the lake...
The lateral stability was much better, when manually testing it by shifting the ship left and right.
Launch...
Swim!


This was too much of the good thing. The ship went off with absolutely no oscillations, it seemed that it sits on on rails. Finally it crashed to the shoreline after having drawn a very wide circle. The crashpoint was 2 meters away from a russian-immigrant fisherman, who is always there when i had unlucky experiences. I like his comments. "Ahh, new ship ey!", "Ahh, better ship now (grin)". The propeller managed to wrap the end of a small branch around its shaft two meters away from the shoreline. I took off my trousers and waded into the swamp to get it out of the water. Must have been a curious looking picture, because i went to the lake on my way back from the office: A man standing in his undertrousers in a lake with shirt and tie, holding a catamarane in his hands, discussing with a russian-based fisherman about fishing spots.

As a next step, i stripped the fins to one third of their original size. And now it was perfect.
No oscillations, straight line...

The two fin sizes together ( i re-glued the cutted segment to see the difference)

What a difference some square-centimeters make!
The new Data:
current: 5.2A @ 11.5V
speed: up to 8 km/h

Thats all till now, time-lapse is over, with episode 11 we have reached the real-time.

Let´s see what the future will bring!

What about starting a colaborating project for UAVs (Unmanned Aquatic Vehicles) on this site?
(The start will be the finding of a better abbreviation...)

Please comment.

E-mail me when people leave their comments –

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

Join diydrones

Comments

  • Hi together,

    thanks for the feedback and hints.
    @michael
    I have to think about the approach with the turn-rate (limiter) and the cross-track error. If i have enough time i will try to realize it.

    Anyway, for my very first objective (survey of a lake / depth profiling) the actual solution is absolutely sufficient. One basic design concept was to keep it as simple as possible and i wanted to have GPS as the only refererence. When starting, i thought, that is absolutely impossible, to get this result. I was amazed, that it worked
    with that accuracy wth GPS and with a simple PI control loop. In older blogs i have posted a google-earth plot with the results. I had cross-track errors in the range of 2-3m on a journey of some 100m, with 180° turns.
    Oscillations only occur, when the ship´s geometry is not ideal (as it was on the firebrigade ship and the cat without fins). For this assemblies, the stabilization with Gyros and/or compass is indispensable. In the first days (before the ArduPilot) i also tried to get the Body-Board assembly stabilized with a heli-gyro, with no effect, because this gyro has other parameters as needed for a ship.

    @SgtRic
    Unbelievable, but it works with this assembly. I also tried "conventional" rudder, but returned to the windmill for ease of construction. The whole assembly can easily moved to another platform for tests (some gaffer tape is sufficient :).
    The trick is to use a ship construction with good lateral stability (the fins help a lot). Even the sloppy attachment of the servo is no problem.
    I will post a video in the next days (it is still raining...)
  • Moderator
    I personally love your blog and enjoy your resourcefulness and experimentation... that is how our community grows...actual hands on trial and error.

    I do wonder about your swivelling motor mount. I have seen other attempts fail because of the adverse thrust line and inbalance this creates.
    Most traditional air-boat designs use a static engine mount and a single or twin air rudder further back on the hull, so as to direct the thrust in a more balanced manner.
  • Another thing we added to the steering routine was cross track error. When the craft reaches it's waypoint, it calculates the bearing to the next waypoint. That bearing is stored and referenced as the craft is traveling.

    Say for arguments sake the heading from WP1 to WP2 is 90 degrees. Now lets say we just made the turn and our actual bearing to target is 1.2 degrees. Lets also say that the distance to the next target is 300 meters. Now using some trig we can find the distance to where we should be on the line between the two waypoints.

    First we get the reverse bearing between the two Waypoints which in this case would be 270. Then using the distance to target we get from our current position to WP2 we throw that value into this formula:

    R = 6378.135
    latitude = Deg(ArcSin(Sin(Rad(Lat1)) * Cos(Distance / R) + Cos(Rad(Lat1)) * Sin(Distance / R) * Cos(Rad(Bearing))))
    longitude = lon1 + Deg(Atan2(Sin(Rad(Bearing)) * Sin(Distance / R) * Cos(Rad(Lat1)), Cos(Distance / R) - Sin(Rad(Lat1)) * Sin(Rad(latitude))))

    lat1=WP2's latitude
    Lon1=WP2's Longitude
    Distance= The current distance that the craft is from WP2. graphically, if you used a drafting compass and drew a circle with WP2 being the center, The line would intersect your current position and the line of position (LOP - where you should be on the line between the two WPs)

    The above equation gives us the Coordinates of where we should be.

    Now we can easily calculate distance left or right of the line we should be on. Add a third PID loop to the steering control algorithms and you can now direct your craft back to the desired LOP.

    Good Luck!
  • If you really want to have stable steering, you need to have at least one gyro for yaw. Break the steering PID down into 2 sections - Turn Rate and heading. Your heading error PID will dictate how much turn rate to use all the way to a set maximum. Then the Turn rate PID will reference the gyro hundreds of times per second and turn the boat at the desired turn rate until you are on course.

    Example: Say you are 90 degrees off of your heading. The Heading PID will be way in error and send that to the Turn Rate PID. the Turn Rate PID will be told at that point to turn at some High rate. However, you have a set maximum turn rate say 5 degrees per second. So the turn rate is hauled in to 5 degrees per second until the heading error is alot closer to zero at which point the turn rate will decrease and eventually you will be on course. The gyros really stabilize the stearing and give the craft an amazing stability.

    It is too difficult we found to turn on heading error alone as the GPS updates come to slowly.
    stability.it - This website is for sale! - Stability Resources and Information.
    This website is for sale! stability.it is your first and best source for all of the information you’re looking for. From general topics to more of wh…
  • @bigkahuna
    100 points! you have been right.

    @abey
    you are right, water props and rudder are better, but i had some paranoia with waterplants that will screw up the propellers / rudders. Anyway, the small fins do a perfect job now, The video i have posted is without any fins. I will post one with the actual assembly as soon as the rain will stop (maybe next week).
    One big advantage of the air propeller: It chases the swans away. Imagine a swan that gets hit by a quiet boat from behind! The boat will be a wreck afterwards :-)
  • So I guessed right, eh? Other options to keep the boat from sining back and forth: a. Add a long keel under the hull b. use a water prop + rudder

    How about ASV for "Autonomous Surface Vessel"?
  • I can see why a aerodynamics out-of-water propeller is more convenient to build and maintain, but at this stage, wouldn't it make sense to go back to a more traditional rudder design. The rudder could be directly attached to the propeller contraption and I'd expect much cleaner steering (less oscillations, higher control gains, more precise behaviour overall...). Am I missing something?
  • @ryan
    decreasing the gain will bring nothing, the GPS update rate and the latency stand against that. The small fins ar e perfect for that problem. Otherwise a gyro or compass (with higher update rates) shall be used.
    @Brian
    AABUB, hmm no bad idea :-)
  • Moderator
    Autonomous Aqua-Based Unmanned Boat - AABUB as in ('ey bub, what chew up to?) That's what you'd get if that fisherman was a new england angler versus your Russian... :)
  • Developer
    turn your heading controller gain way down! and possibly add some derivative! or do something like this:

    http://diydrones.com/profiles/blogs/improved-heading-controller?xg_...
This reply was deleted.