Abstract The ideas of data mining and work flow reconstruction are introduced into intelligent learning of robots. A way to use the first-order and second-order Markovian methods to solve the problems of analyzing the action sequence of robot and process modeling is presented. After experimenting with simu-robot, the process model extracted from one robot can be not only reconstructed in another robot but improved to get a better control performance.
Received: 06 September 2004
Published: 15 July 2005