Mobile Manipulators' Object Recognition Method Based on Multi-sensor Information Fusion

摘要

Mobile manipulators have attracted a lot of interest recently. This paper presents an object detection method for mobile manipulator system based on multi-sensor information fusion. Firstly,the fusion method of visual and ultrasonic information has been applied to localization of the end-effector. And based on the ultrasonic array of the mobile platform, a fuzzy control method was adopted to achieve its obstacle- avoidance motion. Vision-guided object recognition and localization method is emphasizes on image-preprocessing, including that: gray treatment, the choice of threshold, binarization, erosion, dilation and group color up method. These transformations can produce clear image which could be easily recognized by the control computer. A design methodology for moment invariant recognition-based object detection is proposed. Experimental results demonstrate the validity of the approach.