[[alternative]]Development of Virtual Reality Automatic Vehicle Driving Simulation System

碩士機電工程學系[[abstract]]本論文主要以先進安全車輛的觀點為基礎，研究發展一套應用虛擬實境與模糊控制技術之自動車輛駕駛模擬系統。利用虛擬實境軟體Superscape VRT建構行車動線、車輛物件及紅綠燈物件於虛擬場景中，並透過模糊邏輯控制器來控制場景中的車輛物件，使車輛達到預期的目標，以模擬真實世界中車輛之行車狀況。本研究善用Superscape VRT中所賦予的物件屬性，定義被控制車輛之前、後方與側邊為偵測範圍來感測被控制車輛周邊之交通狀況，求得前車與被控制車輛之相對距離與速度，藉以取代實車上利用昂貴的影像測距設備所求得之相同資訊。其次，導入模糊控制理論，將所設計之模糊邏輯控制器轉換為C語言程式，且編譯成一動態連結函式庫之DLL檔案格式，嵌入至虛擬場景中。透過模糊控制法則演算，可計算出場景中之行車狀況，再將偵測資訊參數回饋至虛擬場景中之被控制車輛，提供車輛判斷的依據達到控制的目的。 本研究之結果，被控制車輛可在車道上依道路速限行駛，假使前方車輛車速過慢但路況不允許超車時則與前車保持安全跟車距離；當側邊偵測範圍無車輛時，被控制車輛將執行自動超車之行為模式；再者車輛偵測到紅燈訊號，被控制車輛將停車等候綠燈並禮讓行人穿越馬路，模擬與真實世界自動車輛駕駛行為效果。[[abstract]]Based on the concept of advanced safety vehicles, this study develops an automatic vehicle driving simulation system using Virtual Reality (VR) and fuzzy control technologies. The author uses Superscape VRT (SVRT), a VR software, to set up traffic routes, vehicles objects and traffic light objects in a virtual scenario. The vehicle objects are then controlled by fuzzy logic controller to simulate the conditions of a driving vehicle in the real world. SVRT contains object attributes that are responsible for detecting the conditions around the controlled vehicle through the front, rear, and side of detecting range. The relative distance and velocity between the preceding vehicle and the controlled one can then be calculated instead of using expensive image processing sensors mounted on a real vehicle. Furthermore, combining with fuzzy control theory, a fuzzy logic controller is designed using C language programs, compiled into the Dynamic Link Library (DLL) format, and embedded into the virtual scene. The detected parameters of the virtual scene mentioned above are then sent to the controlled vehicle to provide the information for the vehicle to take actions through fuzzy control rules. The vehicle is thus controlled automatically. There are three simulated situations in this study. First, the controlled vehicle can be driven automatically to follow the speed limit. Secondly, the controlled vehicle is simulated with another vehicle in front of it: if the speed of the preceding vehicle is too slow and the traffic situations do not allow passing, the controlled vehicle will follow the traffic and keep a safe distance between itself and the preceding one. On the other hand, if the speed of preceding vehicle is too slow and there are no other vehicles within the detectable range of the controlled vehicle, the controlled vehicle will overtake the preceding vehicle automatically. Finally, the controlled vehicle would stop before the red light and wait for pedestrians to across. The results show that the developed automatic vehicle driving system can accurately simulate the driving conditions in the real world