主催: The Japan Society of Mechanical Engineers
会議名: ロボティクス・メカトロニクス 講演会2017
開催日: 2017/05/10 - 2017/05/13
This paper proposes a point pair feature (PPF) based pose estimation algorithm. Compared to the traditional PPF approach that searches among selected scene points, our approach is designed to search among all scene points. This is accomplished without increasing the computational time. We propose a pose verification method to improve the speed and recognition rate of the algorithm. We evaluated our approach against synthetic and real scenes. A high recognition rate and speed were demonstrated on 3 industrial objects, which are typically difficult to estimate using the PPF approach. The performance of our method was shown to be superior to traditional PPF approach in terms of recognition rate and speed.