結合光達偵測與模糊控制的自動平行停車系統之設計

本論文提出一種以路邊平行停車為目的的自動停車系統，系統實驗的平台包括了光達以及攝影機，使車輛獲得收集周邊情報的能力，透過馬達控制使車輛能自行找到空車位並執行自動停車程式。此自動停車系統的架構分為兩部分，感知和決策。感知部分會在車輛行駛過程中開始處理感測器所獲取的資訊並進行處理。經過處理後，決策部分會開始通過處理後的資訊確認車輛的位置和車位。確定位置後，根據路徑停車，停入格子後根據傳感器判斷是否需要微調。系統提出的停車程序主要分為三個部分。第一步是找到一個可以停放的空車位。第二步就是利用規劃的路徑進行停車。最後，當車輛不在期望位置內，系統能夠基於模糊控制設計的程式進行微調。綜上所述，此自動駕駛系統可以成功實現平行泊車的目的。

關鍵字

自動泊車；自動駕駛；模糊控制

並列摘要

This thesis presents an intelligent self-driving system for the purpose of automatic parallel parking with obstacle avoidance. This intelligent self-driving vehicle platform contains three major sensors, which are a Web camera, a Zed depth camera and a LiDAR detector. With the information obtained from those sensors, we can acquire very clear circumstances of sidewalk and the distances between our vehicle body and obstacles. The architecture of the automatic parking system can be roughly separated into two parts: sensing and decision-making. The sensing part can start to process the information obtained by the sensors when the vehicle is moving. After appropriate image processing, the decision-making part will start to confirm the location of the vehicle and the empty parking space through the obtained information. After determining the location, the vehicle can be driven according to the desired geometric trajectory, and then determine whether fine tuning is required according to the sensor after parking in the grid. The proposed parallel parking procedure of system is mainly divided into three steps. The first step is to find an empty parking space which is enough for parking. Then, the second step is to use the planning path drawn by the geometric circles for parking. Finally, if the vehicle is not within the desired position, the system is able to use a fine tuning strategy based on fuzzy control design for slight adjustment. In summary, the self-driving system can successfully achieve the purpose of parallel parking with obstacle avoidance, and several practical experiments are given for demonstration.