視覚情報を用いたシーリングロボットの目標軌道の実時間生成法と制御

概要

論文の詳細を見る
In this paper, a method is proposed for generating a desired trajectory of a sealing robot by applying a moving average method to crank points detected by a visual sensor. By the present method, the desired trajectory generated is more robust to noise influence, like stain stuck in the plate to be sealed, than generated by a linear interpolation method. As the trajectory teaching is not needed in the proposed method, all procedure we have to do is inputting the coordinates of starting and end points to the computer. In order to shorten the computation time required for image processing, a parallel processing algorithm for detecting of crank position is developed instead of the series processing method used in the previous papers. A real time visual feedback algorithm is proposed to control the positions of the sealing nozzle and the visual sensor independently. The experiment of a tracking task of a given sealing object is carried out. From the experimental responses, it is shown that obvious improvements in the tracking performance by applying the moving average method can be obtained in comparison with the results obtained by the linear interpolation method.
社団法人日本機械学会の論文
1998-08-25