3D Reconstruction of L-shaped surfaces using a rotating array of ultrasonic sensors

In this paper, the authors present a novel strategy applied to mechatronics device for reconstructing quite accurately smooth orthogonal-connected surfaces. The novel approach is based on the Fuzzy C-Means (FCM) classification and the RANdom SAmple Consensus (RANSAC) fitting. The system consists of four inexpensive ultrasonic sensors moved in the space by means of a controlled digital motor in order to obtain distance measurements for each shaft position. The automatic use of powerful mathematical tools and the physical analysis of an indicator based on the energy permit the detection of the critical region, where the output data feel the significant effect of a non-specular reflection. Experimental results are discussed demonstrating the applicability and the good results of this new strategy.