Implementation of a Mobile Multi-Target Search System with 3D SLAM and Object Localization in Indoor Environments

Abstract

This paper addresses the problem of the recognition and localization of multiple targets while building a three-dimensional map using 3D SLAM (Simultaneous Localization and Mapping) in indoor environments. Since stationary target search systems have the limitations that the target search is conducted passively, this paper presents an implementation of a multi-target search system with a mobile robot and multiple sensors. The mobile multi-target search system consists of a 3D LiDAR and a depth camera on top of a two-wheel mobile robot. Multiple targets are recognized by YOLO (You Only Look Once), and the relative position between the mobile robot and the recognized target is measured by the depth information obtained by the depth camera. The 3D SLAM is implemented with LeGO-LOAM (Lightweight and Ground Optimized Lidar Odometry and Mapping), and the positions of the recognized multiple targets are described with the 3D map. The mobile multi-target search system was implemented on ROS (Robot Operating System) and tested in multiple indoor environments.