- 學位論文
無人載具之動態避障方法研究
Research on Dynamic Obstacle Avoidance for Unmanned Vehicle
摘要
近年來隨著機器人領域技術的快速興起，順向帶動其周邊各個領域的發展，而無人載具便是其中一個顯著的例子。對於無人載具而言，至關重要的課題便是討論如何使得載具自初始狀態經由一系列的運動規劃，順利到達最初設定的目標狀態。對於一般的靜態環境而言，路徑規劃演算法就能解決上述的命題；然對於動態環境而言，單以路徑規劃演算法處理動態環境便顯得稍有不足。 路徑規劃演算法係採用載具當前的位置來判斷載具是否與障礙物發生碰撞，在面對環境中具有動態障礙物的情形時，重新計算並修正原先規劃好的路徑會讓運算成本變高，而本論文採用一種基於載具、障礙物的當前位姿(Posture)、速度來判定兩者之間是否存在碰撞風險的演算法來達成動態環境下的避障，而不至於因重新探索、更新地圖而耗費運算成本。 實驗與模擬結果顯示，在考慮一個差速輪載具的動態模型時，且環境中存在動態障礙物的情況下，若障礙物的位姿、速度資訊完備，則該演算法能有效降低載具與障礙物之間的碰撞風險。
並列摘要
Owning to the rapid development of the area of robotics, the studies on the related field are stimulated. Unmanned vehicle is one of the most notable example. The most significant issue to unmanned vehicle is how the vehicle move autonomously from initial state to final destination via motion planning and control. It’s not that difficult to deal with navigation problem in static environment, but when it comes to navigation problem with dynamic obstacle, merely relying on path planning is not that sufficient. To deal with the navigation problem with moving obstacles, path planning algorithm keeps searching and updating the path, which may cause the waste of computational resources. In this research, we introduce a motion planning algorithm that takes the current posture and velocity of the vehicle and obstacle(dynamical) into consideration to generate the maneuver of the vehicle that would not collide with the obstacles at some future time. Simulations and experiments show that with the implementation of the motion planning algorithm and with adequate information of moving obstacle, a differential drive mobile robot (DDMR) with two caster wheels is capable of avoiding moving obstacle during the navigation and control process.