利用GPS定位及路徑規劃達成電動車於封閉場域內自動駕駛

本論文提出一種利用全球定位系統（GPS）和路徑規劃的自動駕駛路徑追蹤方法，該方法由路徑規劃和路徑追蹤兩部分組成，透過A*搜索演算法搭配搜索限制條件得出符合交通規則之規劃路徑，再以純追蹤演算法與車道維持系統計算車輛在追蹤該路徑時的前輪轉向角度，以及依據路徑資訊進行車速控制。實驗平台為一輛由NVIDIA Jetson TX2嵌入式運算裝置為控制器來控制車速、煞車、EPS的電動車，並且使用GPS模組、鏡頭、六軸IMU來定位車輛位置、影像辨識、計算車輛航向角度。實驗結果中，路徑規劃所得出之路徑皆能符合交通規則，行駛過程中除去一次測試時車輛於轉彎過程中超出車道，其餘測試皆能行駛於正確之車道內，而路徑追蹤過程中直線行駛時最大橫向誤差約為0.5m，轉彎過程時最大橫向誤差約為2.5m。

關鍵字

路徑規劃；影像辨識；路徑追蹤；車道維持；自動駕駛

並列摘要

In this Thesis, a route tracking approach based on global positioning system (GPS) and path planning is proposed for autonomous vehicle. The method is composed of two parts, namely, path planning and route tracking. In path planning stage, the A* search algorithm is used to plan the paths that conform to legal traffic rules. For the second stage, the route tracking, using pure pursuit algorithm and lane keeping system, is utilized to adjust the steering angle of the front wheels. The experimental apparatus implemented in this thesis, is an electric vehicle with NVIDIA Jetson TX2 embedded computing device. Such a control device is designed to control the vehicle motor speed, brakes and electrical power steering (EPS). It is equipped with two GPS modules, a camera and a 6-axis Inertial measurement unit (IMU) sensor. In the experimental results, it is shown both the shortest path planning and the tracking performance are satisfactory and then can validate the proposed algorithms.