近年來無人機產業蓬勃發展逐漸應用在生活當中，例如：搜救任務、運送物資或者是橋樑檢測等，因此無人機的避障系統在執行飛行任務上是未來無人機發展的重要關鍵之一。本研究以無人機在路徑規劃模式下之避障系統研究為目標。
在本研究中首先利用雷達感測器(RPLidar)來收集數據，將數據進行特徵擷取處理後，透過決策樹演算法建立無人機避障系統，來判斷無人機是否要執行避障？當無人機在執行飛行任務的過程中，無人機與障礙物的距離值觸發到避障條件時會啟動避障系統，此時樹莓派會透過MAVLink通訊協議的方式與飛行控制器進行連接，並且傳送訊號即時改變無人機的飛行姿態以及飛行模式進行無人機的避障，最後將此避障系統套用在無人機的路徑規劃模式下，並成功完成實際飛行。
最終飛行測試結果，無人機在路徑規劃模式下能夠離障礙物7公尺的位置成功啟動避障系統，並且無人機會懸停在離障礙物5公尺的位置，同時無人機會依序進行向右、向左、向上三種不同迴避方式，在迴避過程中感測器也會不斷偵測前方是否有障礙物，當成功迴避障礙物後，無人機將會自動完成後續的飛行任務。

In recent years, the drone industry has experienced significant growth and has found applications in various aspects of daily life, such as rescue missions, supply delivery, and bridge inspections. Consequently, the development of obstacle avoidance systems for drones has become crucial for future advancement. The thesis aims to construct an obstacle avoidance system for the drone in Auto mode.
The thesis will establish an obstacle avoidance system using a Decision Tree algorithm. The data collected by the RPLidar sensor is calculated by the algorithm to improve the accuracy of the obstacle avoidance system. During a flight mission, the obstacle avoidance system is activated when the distance between the drone and an obstacle satisfies the predefined condition triggering obstacle avoidance.
　　The connection between the Raspberry Pi and the flight controller is established using the MAVLink communication protocol, allowing real-time modification of the drone's flight attitude and flight mode. This enables the complete development of the obstacle avoidance system for drones operating in Auto mode. Subsequently, the successful implementation of the obstacle avoidance system is verified through actual experiments.
The final flight test results demonstrate that the drones can effectively activate the obstacle avoidance system when the drone is positioned 7 meters away from an obstacle, ensuring avoidance. The drones employ a sequential approach, utilizing three different avoidance methods: right, left, and up. Throughout the avoidance process, the sensor continuously detects the presence of obstacles ahead. Upon successfully avoiding the obstacle, the drones autonomously resume their subsequent flight mission.

[1] E. Dourado, R. Hagemann and A. Thierer, “Operation and Certification of Small Unmanned Aircraft Systems,” Public Interest Comment, Mercatus Center at George Mason University, Arlington, VA, 2015.
[2] Zhuofan Xu, Ruixuan Wei, Qirui Zhang, Kai Zhou and Renke He, “Obstacle Avoidance Algorithm for UAVs in Unknown Environment based on Distributional Perception and Decision Making,” Proceedings of the IEEE Chinese Guidance, Navigation and Control Conference, Aug. 12-14, Nanjing, 2016.
[3] Minjie Zhu and Samuel Cheung, “Autonomous Collision Avoidance for Small Unmanned Aerial Vehicles,” University of California, Davis, 2015.
[4] C. H.　Wu, S. H.　Tu, S. W.　Tu, L. H.　Wang and W. H.　Chen, “Realization of Remote Monitoring and Navigation System for Multiple UAV Swarm Missions: Using 4G/WiFi-Mesh Communications and RTK GPS Positioning Technology,” 2022 International Automatic Control Conference, Nov. 03-06, Kaohsiung Taiwan, 2022.
[5] Nikolaos Baras, Georgios Nantzios, Dimitris Ziouzios and Minas Dasygenis, “Autonomous Obstacle Avoidance Vehicle using LIDAR and an Embedded System,” 2019 International Conference on Modern Circuits and Systems Technologies, May 13-15, Thessaloniki, Greece, 2019.
[6] 陳俊維, 應用決策樹演算法於四旋翼防撞系統設計與驗證, 國立成功大學航空太空工程研究所碩士論文, 2016.
[7] 廖悟暉, 應用支援向量機於四旋翼機避障系統, 國立成功大學航空太空工程研究所論文, 2021.
[8] 柯俊伊, 應用增強式學習模糊控制之四旋翼機避障系統, 國立成功大學航空太空工程研究所論文, 2017.
[9] L. Breiman, J. H. Friedman, C, J. Stone and R. A, Olshen, “Classification and Regression Trees,” Wadsworth, Inc., Monterey, California, 1984.
[10] Daniel Westreich, Justin Lessler and Michele Jonsson Funk, “Propensity score estimation: neural networks, support vector machines, decision trees (CART), and meta-classifiers as alternatives to logistic regression,” Journal of Clinical Epidemiology, vol.63, Iss.8, August, pp.826-833, 2009.
[11] Pang Yong, Li Zengyuan, Chen Erxue, Sun Guoqing, “Lidar Remote Sensing Technology and Its Application in Forestry. Scientia Silvae Sinicae,” Scientia Silvae Sinicae, vol.41, Iss.3, October, pp.129-136, 2005.
[12] Yongsoon Yoon, Tokson Choe, Yongwoon Park, “Obstacle Avoidance for Wheeled Robots in Unknown Environments using Model Predictive Control,” IFAC Proceedings Volumes, vol.41, Iss.2, pp.6792-6797, 2008.
[13] Huaining Sun, Xuegang Hu, “Attribute Selection for Decision Tree Learning with Class Constraint,” Chemometrics and Intelligent Laboratory Systems, vol.163, April, pp.16-23, 2017.
[14] 吳秉勳, 四旋翼防撞系統設計與驗證研究, 國立成功大學航空太空工程研究所論文, 2015.
[15] Xin Zhong Peng, Huei Yung Lin, Jyun Min Dai, “Path Planning and Obstacle Avoidance for Vision Guided Quadrotor UAV Navigation,” 2016 IEEE international Conference on Control and Automation, July 07, Kathmandu, Nepal, 2016.
[16] Huei Yung Lin, Xin Zhong Peng, “Autonomous Quadrotor Navigation with Vision Based Obstacle Avoidance and Path Planning, ” IEEE Access, vol.9, July 19, pp.102450-102459, 2021.
[17] Mehmet Karahan, Cosku Kasnakoglu, “Path Planning and Collision Avoidance with Artificial Intelligence for a Quadrotor UAV,” 2021 International Conference Automatics and Informatics, September 30-October 02, Varna, Bulgaria, 2021.
[18] Pixhawk,https://docs.px4.io/v1.12/en/flight_controller/
[19] RPLidar A1, https://www.slamtec.com/cn/Lidar/A1
[20] Mission Planner, http://ardupilot.org/planner/
[21] Python程式設計, https://github.com/dronekit/dronekit-python
[22] 樹莓派與Pixhawk連線教學, https://rmotex.blogspot.com/2017/10/raspberry-pi-3-pixhawk-mavlinkmacos.html