基於ROS整合SLAM架構於移動機器人室內環境建圖與導航控制

Indoor Environment Mapping and Navigation Control of Mobile Robot Based on ROS Integrated SLAM Architecture

建圖與定位是機器人、自動駕駛、及其他相關自主系統中不可或缺的重要技術。從導航系統的路徑規劃，到無人機的航道繪製，甚至是智能機器人的自主運行，這兩大技術都扮演著核心的角色。SLAM（Simultaneous Localization and Mapping）就是將這兩者結合的技術，它不僅讓機器人能在未知環境中自主導航，還推動許多現代科技的發展與進步，展現了未來無人化技術的巨大潛力。SLAM是一種定位與地圖構建的方法，它能夠在未知環境中，透過一個移動裝置擷取周圍環境數據，並根據數據構建地圖的同時，確定裝置在地圖中的位置。因現今科技的發展，其硬體成本以及核心處理器的入手門檻大大的降低，同時這也代表著SLAM技術逐漸走向大眾親民化。
本次實驗將利用進階精簡指令集機器（Advanced RISC Machine, ARM）架構的Jetson Nano，並搭配機器人作業系統軟體為基底的履帶型自走車平台來做建圖與定位的實驗。SLAM演算法則使用Gmapping並透過修改其中演算法裡的參數找出參數組合。實驗結果呈現在5個不同場地所得到的數據得出在粒子數為50的基準下，建圖的結果明顯優於對照的粒子數設定，同時在定位實驗中也證實了建圖的準確率是會影響到定位與導航的結果。
本次研究結果也能為其他平台提供SLAM技術在資源有限的硬體平台上的應用，擴大其在各種領域的應用範疇。

Mapping and localization are indispensable technologies in robots, autonomous driving, and other related autonomous systems. From the route planning of navigation systems and the flight path plotting of drones to the autonomous operation of intelligent robots, these two technologies play a central role. SLAM (Simultaneous Localization and Mapping) is a technology that combines these two aspects. It not only enables robots to autonomously navigate in unknown environments but also contributes to the development and progress of many modern technologies, showcasing the vast potential of future unmanned technologies.
SLAM is a method for localization and map construction. It can capture the surrounding environment data through a mobile device in an unknown environment and determine the device’s position in the map while constructing it. Due to the advancements in current technology, the hardware cost and the threshold for accessing core processors have significantly decreased. This also means that SLAM technology is gradually becoming more accessible to the general public.
This paper will use the Jetson Nano based on the Advanced RISC Machine (ARM) architecture and pair it with a tracked mobile platform based on the robot operating system software for mapping and localization experiments. The SLAM algorithm employed is Gmapping, with parameters in the algorithm being adjusted to find the optimal combination. The mapping results from experiments in five different venues indicate that, with a base of 50 particles, the mapping outcomes were significantly better than the comparative particle number settings. At the same time, the localization experiments also confirmed that the accuracy of mapping affects the results of localization and navigation.
The results of this study can also offer insights into the application of SLAM technology on resource-limited hardware platforms, expanding its range of applications across various domains.