標題: | 以DSP平台實現自動駕駛車輛於車道保持、車道變換及超車控制 An Embedded DSP Lane-keeping, Lane-change, and Overtaking Control System in Autonomous Vehicles |
作者: | 吳依庭 Yi-Ting Wu 吳炳飛 Bing-Fei Wu 電控工程研究所 |
關鍵字: | 自動車輛;智慧型駕駛系統;2-D超車策略;Vehicle automation;embedded driving control;2-dimension overtaking |
公開日期: | 2007 |
摘要: | 在本篇論文中,我們使用TI的eZdspTM F2812數位訊號處理板作為主要的控制核心,以嵌入式系統實現一智慧型駕駛系統,並驗證在我們的車輛平台(TAIWAN iTS-1)。此論文整合視覺系統、嵌入式控制器、車間通訊設備、車內各項動態感測器、方向盤驅動馬達以及油門和煞車致動器等多項系統,以實現此自動駕駛系統,達成車道保持、車道變換及超車的控制策略,並具有車速變化輔助駕駛及防碰撞警示的功能。 車道保持控制主要藉由視覺系統取得前方車道資訊給予操控所需的偏移量,而車道變換控制器則是經由舒適度制定的側向變加速度量來產生平順的變換軌跡,並轉換成方向盤轉角控制以進行開迴路車道變換的策略。而超車控制則是利用車間距離及相對車速判斷安全的超車時間點,並自動轉動方向盤進行超車策略,若是系統判斷為無法安全超車,會降低車速並警告駕駛者以避免與前車之追撞。此外為了模仿人類的駕駛行為,我們在控制器中設計一模糊增益表用來補償原本回授控制器輸出的方向盤控制命令。整套系統已經在快速道路及與市區道路相似環境上測試成功。 In this thesis, the designed intelligent driving system is implemented based on an embedded DSP platform (eZdspTM F2812) and the overall system has been equipped in our vehicle platform, TAIWAN iTS-1. The developed system integrates heterogeneous subsystems on-line such as a real-time vision system, an embedded controller, a vehicle-to-vehicle (V-V) communication system , in-vehicle sensors, the steering wheel (SW) motor and throttle, brake actuators to realize the autonomous driving system which can provide aids in steering and speed controlling to achieve the lane-keeping, lane-change and safety 2-dimension overtaking. The warning will be given if a potential collision occurs. For lane-keeping control, the lane mark can be obtained by the vision system. As to lane-change control, a smooth trajectory can be designed based on a comfort-specified lateral jerk and converted to a reference SW command tracking to achieve an open-loop lane-change maneuver. As for safety overtaking control, the controller rotates the SW to overtake automatically according to the timing which is decided on the relative distance and velocity. If the distance is not safe, the velocity will be decreased for car-following to avoid collision. In order to imitate human driving behavior, a fuzzy gain scheduling strategy is designed into the controller to compensate the feedback controller by way of adapting to the SW command appropriately. The overall system has been examined on the expressway and urban-like roads. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#GT009512503 http://hdl.handle.net/11536/38226 |
顯示於類別: | 畢業論文 |