초록

용어보기논문에서 용어와 풀이말을 자동 추출한 결과로, 시범 서비스 중입니다. AI-Helper

Abstract ▼ AI-Helper

Bin picking is an essential as well as a basic operation in robot-based part assembly. In this paper, we propose a data-driven method for representing 3D object which is targeted to be used in bin picking scenario. The proposed approach is view-based one, in which 3D object is represented by a set o...

Bin picking is an essential as well as a basic operation in robot-based part assembly. In this paper, we propose a data-driven method for representing 3D object which is targeted to be used in bin picking scenario. The proposed approach is view-based one, in which 3D object is represented by a set of views obtained from adaptively sampled viewpoints, i.e. more views are sampled for more probable object poses. In our approach, physics-based collision detection is employed to simulate the object stacking in virtual bin. A lot of collected examples of bin stacking are obtained by dropping objects into virtual bins to construct view histogram. K-means clustering is applied to sample the predetermined number of characteristic views. The proposed view sampling method can be used to extract 2D or 3D shape descriptor and further employed to model based object recognition system especially for solving bin picking problem.

주제어