自動倣い作業ロボットシステムのための断面形状認識

概要

論文の詳細を見る
Tracking of a feature point in profile data of the workpieces joint, using a laser range finder, is indispensable for automating path-tracking tasks, i.e. welding, sealing, deburring, and so on. In this paper, a tracking method is developed that achieves higher work quality and enables looser constraints on the shape of the workpiece. In this method, a profile is recognized by minimizing energy defined by the relationship between measured profile-data points and an articulated object model (AOM), that adjusts to the various changes in profile data by modeling a workpieces corners, surfaces, and interstices between workpieces. Experiments using a workpieces profile data measured with a scanning range finder are shown that the proposed method tracks the feature point robustly, even in case of the complicated changes, such as shape transformation from concave to convex, interstice variation, and shape break-down due to obstacles. Moreover, experiments using simulated profile data confirm that the proposed method is robust towards the environmental changes.
社団法人日本ロボット学会の論文
2000-09-15