以模糊理論為基礎實現智慧型輪椅機器人之自動停車設計

Abstract

機器人研究中，自動停車控制是一項非常關鍵的研究議題，本論文以自行研製的輪椅機器人為例，提出利用視覺輔助與模糊理論決策達到穩健停車控制。此系統具備雷達定位、視覺回授輔助路徑規劃、基於模糊決策(Fuzzy Decision)之軌跡追蹤控制器以及具抗干擾之緊急停走功能，整合了攝影機、雷射測距儀、超音波雷達、光學編碼器等感測器來輔助。雷射測距儀(Laser Range Finder)資料可透過樣板比對(Pattern Matching)進行機器人自我定位，將輪椅機器人導引至停車格前方位置，接著再以攝影機回授之影像邊緣偵測進行停車格搜尋，以最小平方法提供近似二次曲線的路徑規劃，為避免停車位置的偏差，我們動態地調整依據影像所規劃的路徑，並在進入車格後提供停車格中線資訊，為達輪椅機器人穩定平順的控制結果，基於影像回授的資訊，本論文設計模糊決策控制器控制輪椅角度誤差與中心偏移誤差至系統穩定之平衡點。針對動態複雜環境干擾，為達到安全穩健的停車控制效果，亦設計緊急停走(Emergence Stop and Go)功能以防止撞到障礙物或行人之危險，最後以實驗結果驗證以上所述之自動停車功能。

Autonomous parking control has been an important issue of the robotic research. In this study, an autonomous wheelchair robot parking system is proposed with vision assistance and fuzzy logic control. The autonomous parking system is consisted of the laser range finder based positioning, vision based path planning, reference path tracking control and the emergence stop and go. Thus, the sensors such as the vision system, optical encoder, laser range finder and ultrasonic are equipped for the wheelchair robot. The robot position is estimated by the laser range finder with the pattern matching and guided to the parking region. After extracting the parking space by vision feedback, a dynamical reference trajectory for parking system is generated recurrently. For the robust parking control, the emergence stop and go is provided to avoid the danger with the vision feedback while the robot is disturbed by the obstacles. Finally, the proposed autonomous robot parking system is examined by the experimental results.