ポテンシャル場のマッチングによる物体の位置・姿勢推定とその収束特性

概要

論文の詳細を見る
We propose a 3D object pose estimation method based on pattern matching of the potential fields generated from objects. In our recognition method, the potential fields of both an input object and a model object are generated at first. Secondly, the potential field of the model object is translated and rotated to match with that of the input object. The transformation parameters, i. e., the unknown parameters, are estimated by a nonlinear least squares method, and the position and orientation of the input object are estimated. Furthermore, we clarify some convergence characteristics quantitatively. We also show some ranges, where the solution converges to an optimum value, for two kinds of methods in order to generate the potential field.
一般社団法人日本機械学会の論文
1995-07-25