Design of Indoor Security Robot based on Robot Operating System

Abstract

The design and implementation of indoor security robot can well integrate the two fields of indoor navigation and object detection, in order to achieve a more powerful robot system, the development of this project has certain theoretical research significance and practical application value. The project development is completed in ROS (Robot Operating System). The main tools or frameworks used include AMCL (Adaptive Monte Carlo Localization) package, SLAM (Simultaneous Localization and Mapping) algorithm, Darknet deep learning framework, YOLOv3 (You Only Look Once)algorithm, etc. The main development methods include odometer information fusion, coordinate transformation, localization and mapping, path planning, YOLOv3 model training, function package configuration and deployment. Indoor security robot has two main functions: first, it can complete real-time localization, mapping and navigation of indoor environment through sensors such as lidar and camera; Second, object detection is accomplished through USB camera. Through the detailed analysis and research of the functional design of the two modules, the expected function is finally realized, which can meet the daily use needs.