Multi-knowledge Approach for Mobile Robot Identification of a Changing Environment

In this paper, a large environment is divided into sub-areas to enable a robot to apply precise localization technology efficiently in real time. Sub-area features are represented in a feature information system so that conventional machine learning or data mining approaches can be applied to identify the sub-areas. However, conventional representations with a single body of knowledge encounter many problems when the sub-area features are changed. In order to deal with changing environments, the multi-knowledge approach is applied to the identification of environments. Multi-knowledge is extracted from a feature information system by means of multiple reducts (feature sets) so that a robot with multi-knowledge is capable of identifying an environment with even some changing features. A case-study demonstrates that a robot with multi-knowledge can cope better with the identification of an environment with changing features than conventional single body of knowledge.