以多感測器融合建構機器人於戶外環境之智慧型導航

Abstract

本論文主要開發機器人在室外三度空間中運動時機器人本體狀態（body state）估測器，並以此估測器為基礎建立機器人的避障導航演算法。本體狀態估測器中使用了三種位置感測器GPS（Global Position System）、編碼器、和加速規，以及三種姿態感測器磁力計、傾斜儀、和陀螺儀，以無損型卡爾曼濾波器（Unscented Kalman Filter）架構，估測出完整的機器人本體狀態。其後，並以雙輪機器人與履帶機器人兩個不同的實驗平台，進行實驗驗證，探討估測器在室外的估測表現及評估估測器受到各感測器量測值變化所產生之影響。室外避障導航演算法是以前述估測器所計算出機器人的位置與姿態為基礎，配合以超音波陣列達到環境障礙偵測與避障，來規劃出機器人適當的導航機制，來調控機器人運行之前進速度與角速度。避障導航演算法並先以電腦模擬分析方式，探討系統中各參數對導航特性之影響。之後如同位置與姿態估測器之方式，以雙輪機器人與履帶機器人兩個不同的實驗平台，進行導航的實驗驗證。

The main purpose of this paper is to design a body state estimator on mobile robots with NOAA(Navigation with Obstacle Avoidance Algorithm) in three-dimensional space of outdoor environments. The body state estimator is based on unscented Kalman filter theory, using three positional sensors such as GPS(Global Position System), encoders and accelerometers as well as three attitudinal sensors including magnetometers, inclinometers and gyroscopes. We verified its outdoor estimation abilities as well as the effects of the signal variation of each individual sensor on the estimator through experiments on the tracked mobile robot and the two-wheeled mobile robot. The NOAA strategy is based on the position and attitude information which computed by the estimator, and it is to apply obstacle avoidance algorithm by using array of the ultrasonic sensors for environments sensing, while using two control parameters - speed and angular velocity controlling constant - respectively. Last, we simulated the coupling effects between parameters and situation during the navigation period on the NOAA strategy with computer and validated it through experiments on the tracked and two wheeled mobile robots.