Sensor Fusion Based Fuzzy Rules Learning for Humanitarian Mine Detection

摘要

In this paper, a sensor fusion based fuzzy rules for humanitarian demining are presented. A fuzzy learning algorithm for extracting fuzzy fusion rules from experimental data of robot-manipulated ground penetrating radar (GPR) and metal detector (MD) is presented. The inputs to the fuzzy learning algorithm are features extracted from both a GPR and an MD while its output is a set of fuzzy rules. Applying the learnt fuzzy fusion rules and knowing GPR and the MD features of a given scan, it is possible to decide if there is a land mine and its approximate depth underground. The features chosen for this fusion algorithm are the peak amplitude of a processed GPR output signal and the peak value of the cumulative sum of amplitudes of MD output signal for the same scanned area. Experimental test results are presented for verifying the validity of the proposed learnt fuzzy fusion rule base