GPS動態定位演算法與無人載具實驗

Abstract

摘要 本研究之目的為發展具有精準定位與偵測規避障礙物功能的戶外自動導航車系統。該系統包括移動站、參考站以及主控站。移動站含有運動功能的馬達制動系統、定位功能的GPS接收系統和電子羅盤、偵測障礙功能的雙眼相機系統和超音波系統；參考站具有的第二組GPS接收機可和移動站搭配提供更多定位資訊；主控站則利用無線網路，接收移動站的訊息監測載具的行為。 為進行即時GPS差分定位，我們採用LAMBDA(Least-squares Ambiguity Decorrelation Adjustment)演算法，經由降低整數周波值相關性以提高整數搜尋效率，以求得移動站和參考站兩組GPS接收機差分量之整數周波值。並使用卡曼濾波器(Kalman Filter)修正基線向量的估測值，配合電子羅盤提供的載具姿態，得到移動站的動態資訊。 在障礙物偵測部分，我們採用雙眼相機，透過雙眼立體視覺將影像資料解算得前方障礙物位置，並搭配多方向的超音波感測器偵測處於視覺死角或過於接近的障礙物。經實驗驗證，利用上述感測器所提供的障礙物資料，結合GPS定位及避障路徑演算規劃，我們所設計的無人載具系統確能即時避障，並順利抵達終點位置。

Abstract The purpose of this thesis is to develop an unmanned-vehicle system capable of accurate positioning, obstacle detection, and collision avoidance. The whole system consists of three parts: base station, reference station, and moving vehicle. On the moving vehicle, a servo motor system is used to do motion control; ultrasonic sensors and a stereo camera are adopted for obstacle detection; an electronic compass and a GPS receiver are installed for vehicle positioning. On the reference station, another GPS receiver is used to provide more positioning data. The main station can monitor and control the unmanned vehicle. For accurate positioning, a pair of GPS receivers can allow us calculate the vehicle position by Difference technique. By using the LAMBDA method (Least-squares Ambiguity Decorrelation Adjustment), we can resolve the integer ambiguity problem in an efficient process. Using Kalman Filter can further improve the performance of positioning, and the kinematic date of the vehicle can be obtained. For obstacle-avoiding, we integrate the stereo camera system to detect and locate the obstacles in front of the vehicle, and ultrasonic sensors to detect nearby obstacles. After detecting the obstacles, a map of obstacles is then built, and a feasible path to the destination can be designed by applying the RRT(Rapidly-Random-Tree) technique. Experimental results show that the design process is sound, and the unmanned vehicle can indeed follow the path to the destination without colliding with obstacles.