自走車運用光學雷達進行動態避障之初步研究

隨著科技的進步，工廠的自走車或是各家車廠開發的自駕車，已經是全球產業發展的共同趨勢。在自走車執行所分配的任務時，工廠中一些障礙物可能會影響任務的執行結果，因此，自走車需要具有自行避開各種障礙物的能力，而自動化的工廠中也會多台自走車同時執行任務，所以自走車間的相互溝通協調也十分重要。本研究透過機器人作業系統操作自走車的運動，並運用物聯網的概念建構自走車系統，讓兩台自走車之間相互傳遞位置訊息，使自走車能夠在執行任務的途中也能得知其他自走車的位置，並能做出提前應變措施。而本研究也運用光學雷達進行地圖建構和路徑規劃，自走車在執行任務時，能夠用光學雷達的掃描功能，加強自走車對自身以及周遭環境的認知，並能夠順利地避開障礙物。本研究設計相關實驗，以驗證此自走車系統在執行指派任務時，能互相溝通協作與避障且順利完成任務。實驗結果顯示，在不同的情境下，兩輛自走車皆能運用相互訊息傳遞順利完成指派任務。本實驗考量影響最終位置誤差的因子有兩個，一個是會車情境，另一個是障礙物種類。實驗結果顯示此兩個因子皆有顯著影響。當沒有障礙物時自走車和目的地的誤差最小，當遇到動態障礙物時誤差最大。在不同會車情境中，兩台自走車皆轉彎會產生最大誤差，直行車皆具有較小誤差。要改善此誤差，必須設定較佳的控制參數。

關鍵字

自走車；機器人作業系統(ROS) ；光學雷達；避障

並列摘要

With the advancement of technology, automated guided vehicle (AGV) of factories or the driver-less cars developed by various car factories have become the common trend of global industrial development. When the AGV performs the assigned task, some obstacles in the factory may affect the final result of the task. Therefore, the AGV needs to have the ability to avoid various obstacles by itself, and AGVs in the automated factory will also perform each task at the same time, the communication and coordination between AGVs is also very important. This research uses the robot operating system to operate AGVs, and uses the concept of the Internet of Things to build the AGV system, and two AGV exchange position information to each other. Therefore, AGVs know the position of other AGV during the execution of the task, and can make contingency measures in advance. In this study, optical radar is also used for map construction and path planning. When the AGV is performing tasks, it can use the scanning function of optical radar to strengthen the AGV to perceive environments and it position, and can avoid obstacles successfully. This study designed related experiments to verify that the AGV system can communicate and cooperate with each other, avoid obstacles and successfully complete tasks when performing assigned tasks. The experimental results show that under different circumstances, both AGV can successfully complete the assigned tasks by using mutual data communications. In this experiment, there are two factors considered to affect the final position error, one is the car-motion, and the other is the type of obstacles. The experimental results show that these two factors have significant effects. The error between the AGV and the destination position is the smallest when there is no obstacle, and the error is the largest when there is a dynamic obstacle in the path. In different traffic situations, both AGV will have the largest error with turning actions, and the AGV with straight motion will have smaller errors. To improve these errors, better control parameters must be found and set.