PDA

View Full Version : [Project] Light Following Robot + Neural Network



Pi Robot
10-09-2009, 07:50 AM
This is not exactly the most exciting robot video in the world, but here it is any way.

I just finished up a little demo regarding the use of a simple artificial neural network (ANN) to control a mobile robot. The demonstration is only meant to introduce the concepts and terminology of neural nets rather than being something particularly useful. Also, this video does not deal with learning in ANN's which is what they are most famous for. That will be the topic of a future posting.

Below the video is a link to my report on how it was done. If you are new to artificial neural networks, I hope you find it useful!


http://www.youtube.com/watch?v=rLOZagQWmas

And here is the link to the report: http://www.pirobot.org/blog/0005/


--patrick

zhanx
10-09-2009, 10:44 AM
it may not be the most exciting video but your link may me start thinking of new stuff for the robot i am building. Do you have any more info on the learning ANN?

Suicidal.Banana
10-09-2009, 01:54 PM
same here, even though you say it may not be very exiting, i must say i love this, keep it up man, next up, make it chase your cats (heat sensor?) hehe

Pi Robot
10-09-2009, 03:40 PM
Glad to see there is some interest in neural nets! Since people have been working on these things since the 1940s, there is literally a ton of material on the web. Perhaps a good place to start for an overview would be Wikipedia. (http://en.wikipedia.org/wiki/Artificial_neural_network) Another good site that has explanations of various learning algorithms is http://www.learnartificialneuralnetworks.com/. And of course, you can just Google "artificial neural networks" and you'll never run out of links.

There are also some really nice machine learning libraries out there. I like Aforge.NET (http://www.aforgenet.com/) (C#) and EmguCV (http://www.emgu.com/) (C#, VB.NET, VC++). EmguCV is based on OpenCV (C++). Both packages have machine learning routines as well as piles of visual processing filters.

--patrick

zhanx
10-09-2009, 06:54 PM
yea when i googled it i lost a bit of sleep last night

Pi Robot
11-21-2009, 08:55 AM
This is a followup to my earlier post describing the use of a simple neural network to control a light following robot. In the original demonstration, the connections between input and output neurons were hard coded with values that were known to steer the robot in the right way. In the current demonstration, the neural network is initialized with random connections and the correct behavior has to be learned.

In the video below, the robot begins with a random 2x2 neural network for controlling the motors based on the values of the two light sensors mounted on the front. A supervised learning algorithm employing the Delta Rule is used to train the network by utilizing a known solution to provide the teaching signals five times per second. At the beginning of the video, you can see that the robot turns away from the light and even goes backward. However, within 10-15 seconds, the network is already sufficiently trained to follow the light beam.

For more information, see http://www.pirobot.org/blog/0006/



http://www.youtube.com/watch?v=aCDbP21JcfY

.

yagiuda
03-10-2011, 08:32 AM
Hey if possible can u give details about the hardware of the robot posted above

djsures
03-10-2011, 12:38 PM
Interesting. That is a very similar method to how my 2d and 3d mapping works. Except I collect more datapoints due to the sweeping sensors. Watch the video of one of my robots navigating autonomously without getting stuck, ever... using that similar method. way to go dude!

YouTube - Omnibot Robot Hack Version 3

Pi Robot
03-15-2011, 06:06 PM
Hey if possible can u give details about the hardware of the robot posted above

It was a while ago on a robot that no longer exists, but I believe I was using the Serializer controller (http://www.trossenrobotics.com/store/p/5196-Robotics-Connection-Serializer-WL.aspx) with a pair of their 7.2V gearhead motors (http://www.roboticsconnection.com/p-51-dc-gearhead-robot-motor.aspx) and two light sensors (http://www.phidgets.com/products.php?product_id=1127) from Phidgets. I was then connecting the Serializer to my desktop using the Bluetooth module. Everything was mounted on an 8" acryllic disk from Tap Plastics.

--patrick

yagiuda
06-21-2011, 01:16 PM
Hey i m to make a final year project
and i wanna make a project on neural networks
can ne1 gv me an idea for real gud,big nd innovative project