具定位校正機制之群組機器人路徑規劃與路徑追蹤

Abstract

本論文提出具定位校正機制之群組機器人路徑追蹤以及路徑規劃演算法，透過融合里程計以及雷射距離感測器的設計，來解決里程計存在累積誤差的問題。在定位方面則是引進蒙地卡羅定位法，以粒子濾波器作為主要架構，藉由雷射資訊進行機器人自主定位，透過定位結果補償移動過程中里程計的誤差，來達到更好的控制效果。在路徑規劃方面，考慮到路徑會隨著機器人的移動而有所改變，基於D*Lite 演算法的基礎，並將演算法推廣到群組，視其他機器人為障礙物，進行迴避。並交由MySQL伺服器進行資料統籌，減少機器人的運算負擔，讓機器人可以專心處理感測器資訊。在路徑追蹤方面，本文使用倒階控制法設計運動學控制器，將路徑規劃的結果當作參考信號，以MCL定位以及里程計混合輸出結果取代過去所使用的里程計定位，以避免里程計的累計誤差。實驗結果表示此方法能使機器人準確運行於事先規劃好的軌跡之上，並且能夠發現群組中其他機器人的存在，並嘗試進行迴避。

This thesis presents path tracking algorithm incorporate localization correction mechanism based a planned path. The existed odometer cumulative error problems can be solved by fusing the odometer readings and localization results. For path planning, we apply the D* Lite algorithm to determine an optimal path and extend its use to multi-robot path planning, where other robots are viewed as obstacles to be avoided. A MySQL server is used to integrate information obtained from odometer readings and localization results by a Monte Carlo Localization (MCL) algorithm to reduce the loading of the robots so that they can concentrate on sensing the environment. As for path tracking, the paper uses backstopping method to design a kinematic controller where the planned path is considered as a reference signal. Experimental results show that the proposed method can control the robot to move along the planned path with good agreement.