NeuroCOLT

Neural Networks and Computational Learning Theory

 

About NeuroCOLT

Papers Archive

1994 1995
1996 1997
1998 1999
2000 2001

Books

info@neurocolt.org

NeuroCOLT Technical Report NC-TR-01-112


2001-112
A Robust Implementation of Autonomous Learning in a Mobile Robot
Wolfgang Maass, Gerald Steinbauer and Roland Koholka

ABSTRACT

We discuss a task for mobile robots that requires fast learning in a completely autonomous fashion. Furthermore we describe a rather inexpensive robot "Oskar" that solves this task. Tasks of a similar nature arise in many application areas where one needs to proceed by trial und error, for example in the area of construction, repair, or rescue. The strategy as well as details of the algorithm QUICK-LEARN that Oskar employs -- in addition to longterm learning via a neural network -- are presented in this article. Further details and videos of the robot are available from http://www.igi.tugraz.at/maass/robotik/oskar.

 

 

 

Download Postscript