TARGET LOCALIZATION BASED ON MULTISENSOR FUSION FOR MOBILE ROBOTS

摘要

Using CCD, ultrasonic and infrared sensors, the authors propose a new hierarchical architecture for multisensor fusion, along with its algorithms for mobile robots. According to the architecture, firstly, an approach to locating targets is proposed when only ultrasonic sensors are available. Secondly, Gabor filter technology is employed to recognize unknown targets using the images provided by the CCD alone. Moreover, on the basis of the polynomial approximation, a method for calculating the coordinates of the ground special points, which are taken as the position descriptions of the recognized target, is presented. Thirdly, when CCD and ultrasonic sensors are available simultaneously, elements at the intersection of the information from CCD and ultrasonic sensors are fused to form reliable position coordinates of the recognized target. Last, the algorithm is tested on our designed and developed mobile robot, CASIA-I.