車輛自動駕駛之模糊控制設計與實現

Abstract

隨著汽車的普及率愈來愈高，智慧型運輸系統逐漸成為熱門的研究話題，而其中車輛的智慧型自動駕駛更是人類長久以來的夢想。本論文的研究目的即為車輛自動駕駛，讓車輛能夠偵測周圍環境的情況，自行判斷操作行駛到目的地。為了要得知車輛與週遭狀況，需要使用各種不同的感測器偵測；而要讓實驗車能夠自動駕駛，必須用不同的致動器操作車輛。因此本研究實際建立了一套自動駕駛系統架構，使用CAN整合雷射掃瞄器、加速度計與電子羅盤等感測器的資訊，傳輸至電腦進行自動駕駛控制的運算，然後將控制命令給予DSP操作方向盤、油門與煞車等致動器，達成車輛的自動駕駛。 傳統的比例控制器有準確性與穩定性不足的缺點，本論文提出使用模糊邏輯控制器來改善之。模糊邏輯控制器的設計是依據人類的直覺操作與專家的經驗推論，因此對於參數的變化有很強的適應性，而且能夠提供類似人類思考模式的控制，可以克服自動駕駛控制的困難。由實驗結果可以發現，模糊邏輯控制器可以改善比例控制器的不準確與震盪過大的缺點，而其中設計較詳細的模糊邏輯控制器，其改善自動駕駛的準確性與穩定性較多。本論文提出之自動駕駛系統，已應用於一實驗車(高爾夫球車)上，並於實際校園環境中完成障礙路徑之自動駕駛測試。

An automatic driving system for intelligent vehicles by detecting the environments through the integration of the laser scanner, accelerometer, and electronic compass has been developed in this thesis. All the steering wheel, the accelerator motor, and the brake of the vehicle have been integrated so that the vehicle can be controlled by a DSP microcontroller through the CAN (controller area network) bus efficiently. Since the traditional P controller presents unreliable performance in real experiments, this thesis further proposed two fuzzy logic controllers, FLC_1 and FLC_2 with different number of fuzzy rules, to improve automatic driving performance. Experimental results indicate that both FLCs improve the accuracy and stability of the autonomous driving system. In addition, the more fuzzy rules are adopted in the FLC design, the better the performance of the real system can be obtained. This automatic driving system realized on a golf car has been tested successfully on complicated paths in NCTU campus to prove the feasibility of the proposed method.