Marker tracking for indirect positioning during fabric manipulation

Abstract

We describe here a marker tracking algorithm for indirect positioning during planar fabric manipulation. Indirect positioning is a unique problem during manipulations of deformable objects. Improving the tracking of position by a robotic system contributes to the dexterous manipulation of deformable objects. To formulate this algorithm, we assessed the movement of a single robotic finger moving one manipulated point on a fabric to one positioned point or marker, to the desired point on a floor. To select an appropriate algorithm, we classified disturbances during the positioning of fabrics. To precisely detect the position of the marker during these disturbances, we applied the combination of a particle filter and a labeling processing to the algorithm. Experimental evidence showed that, due to its precision in detecting position, this algorithm was suitable for indirect positioning. © 2013 Springer Science+Business Media Dordrecht.