[...]... approach More specifically, we shall focus on the problem of sensorless orientation of parts in which no sensory information at all is used to move the part from an unknown initial pose into a unique – and known – final pose In sensorless orientation or part feeding parts are oriented using passive mechanical compliance The input of the problem of sensorless orientation is a description of the shape of the... Roy and Cox [13] use maximumflow computations in graphs to solve this problem, yielding considerably better results, but unfortunately at increased computational costs G.D Hager et al (Eds.): Sensor Based Intelligent Robots, LNCS 2238, pp 25–38, 2002 c Springer-Verlag Berlin Heidelberg 2002 26 Boris Kluge Matching multiple moving point objects by computing minimum cost matchings in bipartite graphs is... direction of motion of the finger is altered and the action is repeated The problem os sensorless orientation by a pulling finger is to find a sequence of motion directions that will cause the finger to move the part from any initial pose into a unique final pose Although intuitively similar to pushing it turns out that sensorless orienting by pull actions is considerably harder than pushing [14] As the... In IEEE International Conference on Robotics and Automation (ICRA), pages 846–851, 1996 25 M A Erdmann and M T Mason An exploration of sensorless manipulation IEEE Journal of Robotics and Automation, 4:367–379, 1988 26 K Y Goldberg Orienting polygonal parts without sensors Algorithmica, 10(2):201–225, 1993 27 M Jakiela and J Krishnasamy Computer simulation of vibratory parts feeding and assembly In... asymmetric polyhedral part such that the part gets oriented on its descent along plates and fences This paper reports on parts of our research in the field of sensorless manipulation of the last few years The emphasis will be on the transformation of various sensorless part feeder problems into geometric problems, a sketch of the algorithms that solve these problems, and on determining classes of orientable... of sensorless orientation is a description of the shape of the part and the ouput is a sequence of open-loop actions that moves the part from an unknown initial pose into a unique final pose Among the sensorless part feeders considered in the literature are the traditional bowl feeder [18,19], the paralleljaw gripper [23,26], the single pushing jaw [3,29,31,34], the conveyor belt with a sequence of... observe that thin parts are easier to orient than fat ones A theoretical analysis confirms this intuition To formalize our intuition about fatness, we define the geometric eccentricity of a planar part based on the lengthto-width ratio of a distinguished type of bounding box We deduce an upper bound on the number of actions required to orient a part that depends only on the eccentricity of the part The... P Note that the minimum eccentricity of 0 is in both our definition and in the definition for ellipses obtained for circles Chen and Ierardi [23] proposed a class of plans for orienting polygonal parts based on repeating a unique push-and-reorient operation The length of the longest angular interval without equilibrium orientation, or, in other words, of the longest half-step of the push function, determines... relatively low even for smaller values of ; N ≈ 75 for = 0.5, N is below 50 for = 1 and below 30 for = 2.5 Similar bounds can be obtained for squeeze plans [37] 2.3 Pulling Parts We have recently studied sensorless orientation of planar parts with elevated edges by inside-out pull actions [14] In a pull action a finger is moved (from the inside of the part) towards the boundary As it reaches the boundary... rigidly attached to both its sides [20,35,39], the conveyor belt with a single rotational fence [2], the tilting tray [25,33], and vibratory plates and programmable vector fields [16,17] Traditionally, sensorless part feeding is accomplished by the vibratory bowl feeder, which is a bowl that is surrounded by a helical metal track and filled with parts [18,19], see Figure 1 The bowl and track undergo an . Kong London Milan Paris Tokyo Gregory D. Hager Henrik I. Christensen Horst Bunke Rolf Klein (Eds.) Sensor Based Intelligent Robots International Workshop Dagstuhl Castle, Germany, October. Data applied for Die Deutsche Bibliothek - CIP-Einheitsaufnahme Sensor based intelligent robots : international workshop, Dagstuhl Castle, Germany, October