my robot fish, revealed!

[Admin]

For almost two years now Ive been asking weird questions on the forum to get help for my big robot fish project. But unfortunately, being a government project, I wasn't allowed to reveal the details . . . until now!

I wrote up a big 8 page science paper on it, and submitted it to ICRA (International Conference on Robotics and Automation). In all, ICRA recieved 1800 papers covering all types of robotics research.

Yet despite this being the first paper I ever wrote, out of that 1800 papers it was given the 'Best Conference Paper Award' - so its bound to be a good read

Without further introduction, here it is (800kb):
http://www.societyofrobots.com/robottheory/Design_of_a_Biomimetic_Controlled-Curvature_Robotic_Pectoral_Fin.pdf

Now after looking at all the pictures you may be asking something like, 'well, where is the fish part?'

Well . . . its going to be made public when I publish the next paper on it at the end of this year

I guess this now gives me some official credibility in robotics now
. . . and shows that the CMU Robotics Institute made a huge mistake in rejecting me

[ed1380]

the design of the fin looks cool. I only wish I understood more of that

[JonHylands]

Wow, just finished the paper - I am really impressed !

I wonder how a flexible articulated tail fin like that would do... I think you would need the articulation on a different axis, but it may be that being able to independently articulate the top and bottom of the fin from the center would produce more thrust.

Very cool stuff, can't wait to see the finished AUV...

- Jon

[sdk32285]

Cool paper

Congrats on becoming published!

[nanob0t]

That is insanely wicked.  It's a pretty neat concept, I only wish I could get into stuff like that right now.  Unfortunately, I'm 16 and the most that a program can offer is playing with LEGO mindstorms, which I own too much at.  I had that at like 10.

It's a very good paper.  I hope to do stuff like this in the future, I'm interested in seeing the next paper.

Yet despite this being the first paper I ever wrote, out of that 1800 papers it was given the 'Best Conference Paper Award' - so its bound to be a good read

Very nice job.   

. . . and shows that the CMU Robotics Institute made a huge mistake in rejecting me

I bet!   

[Admin]

Part 2 of our robot fish project has been published.

The first paper was designing and studying a robot fin. This paper however covers how to control a vehicle with flapping fins. Unlike with traditional propellor thrusters that give predictable steady thrust in a single direction, flapping fins gives crazy thrusts in many different directions. No one has ever figured out how to control a UUV (unmanned underwater vehicle) with pectoral fins that can change shape, until now.

I wouldnt say its a 100% solved problem, but its definitely an advancement.

Warning - being a controls paper, it is heavy in math and derivation. But being a 2nd author on the paper I can promise you nice CAD models of the design and my own personal touch.

http://www.societyofrobots.com/robottheory/Modeling_and_Control_of_Pectoral_Fin_Propelled_UUV.pdf

I am currently in New Hampshire presenting the paper with two of my collegues for the next few days.

Unmanned Untethered Submersible Technology Symposium
http://www.ausi.org/uust/uust.html

Ill take videos and report for everyone

The next paper will be the finished product . . . probably another 6 months or so of research . . .

In other news, we are now competing (and working) with MIT:
http://news.bbc.co.uk/2/hi/technology/6924057.stm
http://web.mit.edu/newsoffice/2007/robofin-0730.html

They take a completely different approach than us, but still interesting . . .

[mpeg2tom]

Looks cool!  You should check out some of pretty robo fish done by artist U-Ram Choe...

http://www.uram.net/english/portfolio_new/portfolio3rd_05.htm

[JesseWelling]

Isn't it kinda crazy that is seems controllable with PID?
I would have thought some crazy calculus + fuzzy logic would be optimal...
I guess that's why I'm putting in my apprenticeship (so to speak) in a control oriented field

[Hal9000]

I'm glad you got it finished and such! Looks really good. well done!

[Admin]

Isn't it kinda crazy that is seems controllable with PID?
I would have thought some crazy calculus + fuzzy logic would be optimal...

Actually, you are right. The CFD does use crazy calculus, and we are implementing the CFD results into the robot controller using fuzzy logic

Our modeling is what uses the PID.

[JesseWelling]

ohhh well maybe i should RTFA 

[crossroads1946]

Bummer. Another one of my ground-breaking ideas has already been taken. Oh well. I guess I won't have to start from scratch when I build mine now. Nice work!

[jklv]

Amazing, it is an honor to be part of your forum

[Admin]

Hmmmm its been almost exactly one year since the last time I posted anything on my robot fish.

Well, got you two more papers:

This one summarizes the project (pectoral fin UUV design, UUV control, and CFD simulations of the UUV) as of ~10 months ago . . .
Design, Development, and Testing of Flapping Fins with Actively Controlled Curvature for an Unmanned Underwater Vehicle

And this one goes into details on UUV control by pectoral fins, using simulation for validation . . .
Fuzzy Logic PID Based Control Design and Performance for a Pectoral Fin Propelled Unmanned Underwater Vehicle


For those who like math and controls, there's lots of it. For those who don't, there are lots of pictures of my robot fish

So why does it take forever to post stuff on my fish? Well, because science is sloooowwwwwww. From inception of a theory to experimentally proving it takes about a year. Then another ~3 months to write a paper on it. Then another ~2-4 months until it is accepted by a journal or conference.

I currently have three more papers in the works, one 95% done, one 60% done, and the last 10% done. I got good results and am mostly finished, but it will take another ~6 months to get it all published.

[sdk32285]

Hey
I like that second paper (not to take away from the first paper, but I am a controls guy).

The idea of having the fuzzy logic stage after the main PID controller is interesting. I'm thinking that I can adapt that to a slow moving wheeled rover. The rover has just two ticks per revolution on each wheel which makes it difficult to implement a smooth constant velocity controller that works while going over obstacles or other "features".

My idea is after the PID stage I can add a "fuzzy" stage where I compensate for changes in acceleration and rover pitch.
(I don't plan on publishing this control work, but if I do Ill reference you)

[Admin]

Ok so its been like a year since I've reported anything new on this project. Well, its because I traveled 8 months around the world (mostly Thailand), so didn't get to work on anything. Below you'll find two new papers recently published. Three more are in the pipeline, hopefully getting published by early next year.

Its heavy reading, lots of controls, CFD stuff, fluid mechanics, etc. I do however strongly recommend looking at the pics. And yeap, pics of the prototype fish swimming around in water!

This following paper models the robot fish in CFD to determine drag coefficients and the effects of the fish body on the fins. This is a 5mb file.

    * Ravi Ramamurti, Jason Geder, John S. Palmisano, Banahalli Ratna, William Sandberg, "Computations of Flapping Fin Propulsion for UUV Design," 47th AIAA Aerospace Sciences Conference, AIAA-2009-0724, Washington, DC, January 2009.

This paper shows how the CFD computed drag coefficients are used in a UUV controls simulation. It also shows the prototype experimental fish (actual pics) doing various preliminary maneuvers.

    * Jason Geder, Ravi Ramamurti, John S. Palmisano, Marius Pruessner, Banahalli Ratna, William Sandberg, "Sensor Data Fusion and Submerged Test Results of a Pectoral Fin Propelled UUV," International Symposium on Unmanned Untethered Submersible Technology, Durham, New Hampshire, USA on August 23rd-26th, 2009.


edit: fixed link

[cuke]

Wow ,it's so good!!

[billblack]

Nice work!
I just saw this in an AUVSI email...Thought you might like to see it if you haven't already:
http://www.flightglobal.com/articles/2009/08/12/330963/auvsi-robot-fish-enters-second-phase-of-development.html