隨著無人搬運車(AGV)在倉儲、物流和製造業等領域的普及，使用機器人在運輸和操作物品方面的效率和安全性受到越來越多的關注。然而，當AGV操作場域複雜或不確定環境時，其運動控制和防碰撞設計仍然存在挑戰。為了實現避障，本論文採用Timed-Elastic-Band (TEB)演算法，在多個選擇路徑中選擇最佳路徑，並使用動態控制策略實現平順移動。此外，針對雷射掃描無法有效偵測之空間障礙物，本論文整合影像辨識來輔助TEB演算法的防碰撞策略，以增強無人搬運車進行導航任務的運動控制和避障能力。透過攝影機即時偵測AGV前方的環境影像，並利用機器學習技術識別空間障礙物的相對位置資訊，透過座標轉換將空間障礙物座標投影在代價地圖上，使TEB局部路徑規劃器可以將其納入計算避障路徑。本論文所開發的防撞策略先於ROS Stage模擬驗證後再將其實現於AGV平台進行實車驗證，透過融合影像偵測資訊與雷射掃描資訊的TEB避障演算法，經由實驗結果驗證能在導航過程中安全完成障礙物閃避。本論文採用之AGV平台及測試場域與業界緊密合作，顯示所提出防撞策略已成功整合於導航軟體架構與實際產業上之需求潛力。

As automated guided vehicles (AGV) become more prevalent in fields such as warehousing, logistics, and manufacturing, their efficiency and safety in handling and transporting goods are becoming increasingly important. However, challenges in terms of motion control and collision avoidance capabilities still exist for AGV while operating in complex or uncertain environments. As such, to enhance the motion control and collision avoidance capabilities of AGV, this thesis proposes an effective strategy based on the Timed-Elastic-Band (TEB) algorithm with the aid of visual recognition. To avoid possible collision, the TEB algorithm is to choose the optimal path from multiple candidate paths and then the dynamic control can be utilized for smooth motion control of AGV. In addition to the obstacles detected from 2D laser scanning, the proposed strategy uses a camera to capture the front environment of the AGV and uses machine learning techniques to recognize the camera coordinate of the spatial obstacle in real time. Thus, such a spatial obstacle can be projected onto a cost map for the TEB path planner to determine the collision-free path. With consideration of the kinematics and dynamics constraints of our AGV, the collision avoidance strategy presented in this study is firstly verified in Ros Stage simulation. Further, the experimental evaluation for automatic navigation process combined with global path planning and TEB local path planner is carried out. Through the TEB obstacle avoidance algorithm fusing visual information and laser scanning information, experimental results verified that obstacles can be safely avoided during the navigation process. The AGV platform and test field used in this thesis are in close cooperation with the industry, showing that the proposed collision avoidance strategy has feasible potential integrated into the navigation software architecture and the demand of actual industry.

[1] M. S. Wiig, K. Y. Pettersen and T. R. Krogstad, "Collision Avoidance for Underactuated Marine Vehicles Using the Constant Avoidance Angle Algorithm," in IEEE Transactions on Control Systems Technology, vol. 28, no. 3, pp. 951-966, May 2020, doi: 10.1109/TCST.2019.2903451.
[2] A. S. Lafmejani, H. Farivarnejad and S. Berman, "Adaptation of Gradient-Based Navigation Control for Holonomic Robots to Nonholonomic Robots," in IEEE Robotics and Automation Letters, vol. 6, no. 1, pp. 191-198, Jan. 2021, doi: 10.1109/LRA.2020.3037855.
[3] 鄭敬錡，”基於ROS開發工業應用之無人搬運車安全及強健移動式機器人導航策略”，國立臺灣師範大學，碩士論文，2020年
[4] Global and Local Path Planning Study in a ROS-Based Research Platform for Autonomous Vehicles https://doi.org/10.1155/2018/6392697
[5] M. B. Alatise and G. P. Hancke, "A Review on Challenges of Autonomous Mobile Robot and Sensor Fusion Methods," in IEEE Access, vol. 8, pp. 39830-39846, 2020, doi: 10.1109/ACCESS.2020.2975643.
[6] Chen, H.; Lu, P. Real-time identification and avoidance of simultaneous static and dynamic obstacles on point cloud for UAVs navigation. Robot. Auton. Syst. 2022, 154, 104124.
[7] F. Rovira-Más, V. Saiz-Rubio and A. Cuenca-Cuenca, "Augmented Perception for Agricultural Robots Navigation," in IEEE Sensors Journal, vol. 21, no. 10, pp. 11712-11727, 15 May15, 2021, doi: 10.1109/JSEN.2020.3016081.
[8] G. Chen, H. Yu, W. Dong, X. Sheng, X. Zhu and H. Ding, "What should be the input: Investigating the environment representations in sim-to-real transfer for navigation tasks", Robot. Auton. Syst., vol. 153, 2022.
[9] Y. Zhang, G. Tian, X. Shao and J. Cheng, "Effective Safety Strategy for Mobile Robots Based on Laser-Visual Fusion in Home Environments," in IEEE Transactions on Systems, Man, and Cybernetics: Systems, vol. 52, no. 7, pp. 4138-4150, July 2022, doi: 10.1109/TSMC.2021.3090443.
[10] 王思涵，”可適應無人搬運車彈性化設計之學習式導航策略及強健式路徑跟隨控制”， 國立臺灣師範大學，碩士論文，2022年
[11] Logitech C922 PRO HD STREAM WEBCAM，https://www.logitech.com/zh
tw/products/webcams/c922-pro-stream-webcam.960-001091.html
[12] 機器人作業系統 ROS，取自 https://www.ros.org/
[13] GMapping，取自 http://wiki.ros.org/gmapping
[14] AMCL，取自 http://wiki.ros.org/amcl[15] Move Base，取自http://wiki.ros.org/move_base
[16] Move Base 完全詳解，取自 https://blog.csdn.net/u013834525/article/details/84627204
[17] Costmap， 取自 http://wiki.ros.org/costmap_2d
[18] C. Roesmann, W. Feiten, T. Woesch, F. Hoffmann and T. Bertram, "Trajectory modification considering dynamic constraints of autonomous robots," ROBOTIK 2012; 7th German Conference on Robotics, Munich, Germany, 2012, pp. 1-6.
[19] MediaPipe，取自
https://github.com/google/mediapipe/blob/master/docs/solutions/objectron.md
[20] ROS Stage，取自 http://wiki.ros.org/stage
[21] O. Khatib, “Real-time obstacle avoidance for manipulators and mobile robots,” in Proc. IEEE Int. Conf. Robot. Autom., 1985, pp. 500–505.
[22] E. Rimon and D. E. Koditschek, “Exact robot navigation using artificial potential functions,” IEEE Trans. Robot. Autom., vol. 8, no. 5, pp. 501–518, Oct. 1992.
[23] Vaughan, R.T., Gerkey, B.P. (2007). Reusable Robot Software and the Player/Stage Project. In: Brugali, D. (eds) Software Engineering for Experimental Robotics. Springer Tracts in Advanced Robotics, vol 30. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540- 68951-5_16