Title :
Learning in robotics by experimentation
Author_Institution :
Fac. of Electr. Eng. & Comput. Sci., Ljubljana Univ., Yugoslavia
Abstract :
The possibility of a robot, as an autonomous agent, learning through experimentation is investigated. A robot is embedded in an environment that can be sensed and acted upon. Through experimentation a robot explores the domain of object pushing and by using techniques of machine learning induces the relationship between its action and perception. The rules generated through the process of inductive learning are represented in the form of a qualitative model
Keywords :
learning systems; robots; experimentation; inductive learning; machine learning; object pushing; qualitative model; robot; robotics; rules; Autonomous agents; Computer science; Hardware; Humans; Machine learning; Machinery; Physics; Robot programming; Robot sensing systems; Robustness;
Conference_Titel :
Electrotechnical Conference, 1991. Proceedings., 6th Mediterranean
Conference_Location :
LJubljana
Print_ISBN :
0-87942-655-1
DOI :
10.1109/MELCON.1991.162046
