Neural Network Manager for the NXT
Christoph Bartneck, Assistent Professor at the Eindhoven University of Technology, has blogged about a new Java-based application with a graphical user interface that "allows Lego fans to easily create, manage and train a neural network which can be exported and used in any program to be designed to run on the LEGO NXT."
From the web site:
"The total system comprises of two main parts, a UI running on the NXT and the Neural Network Manager running on the PC. The Neural network is created by the Neural Network Manager and can once trained be exported to a Java file that should be compiled and uploaded to the NXT."Now, doesn't that sound interesting?
Mainly to deduce functions/algorithms from observations (sensor data) where the input data are very complex and thus such a function or algorithmus cannot be derived a priori.
That's the case for the environment living creatures are moving in, for instance, so their brains are very complex neural networks (to say it somewhat simplified) that learn (are trained) from the beginning of their existence on.
Hence, neural networks can be very important tools in robotics in training autonomous robots to move around in complex environments: real adaptive robots that behave appropriately in a complex and changing environment must be able to make decisions of their own (that are not pre-programmed) and to learn from former experiences.
it's assistAnt, not assistent.
Thanks in all cases