View unanswered posts | View active topics It is currently Wed Dec 17, 2014 6:02 pm






Reply to topic  [ 3 posts ] 
NXT Copter with Optimal Control 
Author Message
Rookie
User avatar

Joined: Fri Mar 21, 2008 8:00 pm
Posts: 20
Location: Spain
Post NXT Copter with Optimal Control
Hi,

This is my last Project, where I've developed a new technology (CACM-RL) to control Nonlinear and Unstable Robots, like aircrafts, Xcopters, cars, anything, without any mathematical model, because I've applied auto-learning to autodiscover the platform and Optimal Control to obtain the Optimal Controller... obviously.



http://youtu.be/cKHisPFz56U

I've used NXT to develop this demo, but I will like someone knock to my door with a interesting challeng and money.

I hope you like, and I appreciate any support...

:-)

_________________
Nobody knows what you're capable ... until you get down to it.


Fri Nov 01, 2013 11:31 am
Profile WWW
Senior Roboticist
User avatar

Joined: Wed Sep 28, 2011 10:13 pm
Posts: 630
Location: If I told you, I'd have to kill you.
Post Re: NXT Copter with Optimal Control
That is quite interesting! Very smooth and precise control! Any chance of seeing the source code?

_________________
A.K.A. inxt-generation
Self-proclaimed genius, and future world dominator.
My Brickshelf Folder
"Don't they teach recreational mathematics anymore?" - The Tenth Doctor
Bow down to Nikola Tesla, King of the Geek Gods.


Fri Nov 01, 2013 11:45 am
Profile WWW
Rookie
User avatar

Joined: Fri Mar 21, 2008 8:00 pm
Posts: 20
Location: Spain
Post Re: NXT Copter with Optimal Control
The algorithm is under Patent, but the basic idea is to estimate the vector state (Th, Vth), and look into a Optimal Control Table, the optimal action associated to this state, iteratively.

In this way the NXT computation is very light, because it only has to look into a memory cell addressed by the status vector.

The dificulty is to construct the Optimal Control Table. It's a checkerboard where each state has one action.... the optimal, and the algoritm to construct this table is based in Reinforcement Learning, to avoid in this way to have to work with a math-model.

_________________
Nobody knows what you're capable ... until you get down to it.


Fri Nov 01, 2013 12:03 pm
Profile WWW
Display posts from previous:  Sort by  
Reply to topic   [ 3 posts ] 

Who is online

Users browsing this forum: No registered users and 2 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  



Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group.
Designed by ST Software for PTF.