Humanoid Robots for Simultaneous Localization and Mapping Based on Robot Operating System

[[abstract]]定位是機器人在人工智慧中一項不可或缺的功能。本研究使用單個攝影機作為機器人的電腦視覺，因此在仿人形機器人系統上自我定位以及地圖建立是一大挑戰。 本論文設計蒙地卡羅定位法來實現機器人自我定位並同時偵測障礙物建立至地圖中。為了增加收斂速度與準確率，在機器人上加裝了陀螺儀來改善定位的效果。在定位的過程中運算量相當大，所以本研究架構建立在機器人作業系統(Robot Operating System；ROS)上。機器人作業系統主要為分散式架構，以點對點方式將所有處理程序相互連接，因此可使系統達到平行處理，此外還有相當完善的可視化工具，進而使得機器人能夠順利的在環境中順利的自我定位並建立地圖。[[abstract]]Localization is the fundamental work to develop the artificial intelligence of robot In this research humanoid robot vision only using a webcam is a big challenge of Simultaneous Localization And Mapping (SLAM) In this thesis Monte Carlo localization method for simultaneous localization and update the map information immediately for unknown obstacles In order to increase the convergence speed and accuracy using IMU (Inertia Motion Unit) improving the effect for the localization error Because of a large amount of computation to locate the structure of experiment is built on Robot Operating System (ROS) for running efficiency Since being the distributed architecture ROS handles all procedures by peer-to-peer method that can allow the system to achieve parallel processing In addition it also has a very comprehensive visualization tools so that users can be easy to develop SLAM for robot