During the multi-source information fusion with Dezert-Smarandache theory（D-SmT）

the problem of the subjectivity exists in getting the mathematical model of general basic belief assignment（GBBA）.A new objective method is obtained which is integrated with rough ret（RS）theory.The dependency degree among data attributions is identified by analyzing the massive data collected by the sensors.With the concept of classification and coordinated decision system

the data is discreted to inform decision tables.The concept of rule intensity is adopted to build GBBA model which would be stored in database to be queried when necessary.P2-DX robot is used as a platform and sonar sensors on the mobile robot are introduced as example.The correctness and validity of the developed method are verified according to sonar’s electrical characteristic.