Object Localization Using Light Beam Sensing

We implemented a localization method which partially solves the problems of placement and assembly. The idea was to localize objects precisely (1 mil) so that uncertainty was negligible and dead reckoned positions were sufficient. The method was g eared towards industry by using reliable accurate, robust, inexpensive, sensors; light beam sensors have already been used in manufacturing environments. This technique involved passing objects through a set of coplanar beams and recording the robot's position when a light beam was broken or a beam reconnected. The algorithm localized and identified objects with the assumption that they were rigidly held. The advantages of this approach were: the feasibility and precision of the sensors, constant time identification using precomputed hash tables, and worst-case linear time identification algorithm. The disadvantages were: it cannot distinguish objects with equivalent convex-hulls, and does not work on "thin" objects. The performance and limitations of the actual beam sensors affected the design of the apparatus