勢能輔助速度障區動態避障策略設計與實驗

Abstract

本研究以勢能法輔助速度障區完成避障任務，以改良式速度障區算法獲得非碰撞區域，並由人工勢能場輔助找出非碰撞區中梯度最低的一條路徑。在實驗設計部分，結合傳感器實現自主定位，獲得位置、姿態、速度等資訊，再以模糊控制器計算對差速輪載具所需要輸出的左右輪轉速完成任務。 模擬與實驗結果顯示，加入勢能場修正了速度障區所計算出的路徑，使得選擇到的路徑能與障礙物保持適當距離，在複雜環境中更容易找出一條遠離可能碰撞區域的路線。在實驗部分，本研究使用視覺感測器及卡爾曼濾波器進行自主定位，再利用光達及視覺感測器判定障礙物的位置及行進方向，使得本研究所發展的避障策略得以順利完成任務。

In this research, an Artificial Potential Field algorithm is designed to assist modified Velocity Obstacle algorithm to accomplish the tasks of obstacle avoidance. The non-collision zones are first obtained using the Velocity Obstacle algorithm, and then, with the aid of the Artificial Potential Field, the path with the lowest gradient in the non-collision zone is obtained. In the experimental work, various sensors are integrated to obtain information such as velocity, orientation, position, and other relevant data. Fuzzy controller is employed to calculate the required speeds of the left and the right wheels of a differential-drive vehicle to complete the task. Simulation and experimental results show that incorporating the potential field method corrects the path computed by the modified Velocity Obstacle algorithm, ensuring that the selected path maintains a suitable distance from the obstacles. A route that stays away from potentially collision-prone areas in complex environments is designed. In the experiments, the visual sensors and Kalman filters are used for autonomous localization, while lidar and visual sensors are employed to determine the positions and the direction of motion of the moving obstacles. From the experimental results, it is seen that the developed obstacle avoidance strategy enables the mission be successful fulfilled.