Development of a Person Search Components Based on an Omnidirectional Camera and a 3D-LiDAR

摘要

In this paper, we propose a method to search the specified person using a spherical camera and a 3D range sensor. This is one of the items of the Tsukuba Challenge. By using the camera and the sensor, the view field becomes wider than the conventional system. We employed color information to reduce the computational load in image processing. In addition, we focused on the view size of the target person to improve the accuracy of LiDAR. The proposed component is implemented as a ROS-based application, which allows for rapid verification as well as rapid prototyping. The validity of the proposed algorithm was confirmed by preliminary experiments simulating the Tsukuba challenge environment.