engleski

Recognition of Objects and Places in 3D Point Clouds for Robotic Applications

In this paper, object and place recognition are considered as the same problem with two applications in robotics: recognition of objects for robot manipulation and recognition of places for mobile robot localization. The approaches which use 3D point clouds obtained by 3D sensors as input are discussed. The paper focuses on feature-based recognition methods which are suitable for robotic applications since they provide a precise pose of either an object of interest or a mobile robot in its environment. Several state-of-the-art object and place recognition approaches are reviewed. An indoor place recognition system designed by the authors is presented as an example.

computer vision; object recognition; place recognition; 3D sensors

nije evidentirano