Using partial sensor information to orient parts

Parts orienting the process of bringing parts in initially unknown orientat ions to a goal orientation, is an important aspect of automated assembly. T he most common industrial orienting systems are vibratory bowl feeders, whi ch use the shape and mass properties of parts to orient them. Bowl feeders rely on a sequence of mechanical operations and typically do not use sensor s. In this paper, we describe the use of partial information sensors along with a sequence of pushing operations to eliminate uncertainty in the orien tations of parts. We characterize the shorter execution lengths of sensor-b ased plans and show that sensor-based plans are more powerful than sensorle ss plans in that they can bring a larger class of parts to distinct orienta tions. We characterize the relation among part shape, orientability, and recogniza bility to identify conditions under which a single plan can orient and reco gnize multiple part shapes. Although part shape determines the results of t he actions and the sensed information, we establish that differences in par t shape do not always lead to differences in part behavior. We show that fo r any convex polygon, there exists an infinite set of nonsimilar convex pol ygons that behave identically under linear normal pushes. Furthermore, ther e exists an infinite set of nonsimilar convex polygons whose behavior canno t be distinguished even with diameter sensing after each push. We have impl emented several planners and demonstrated generated plans in experiments.