中文摘要
多功能智慧行車輔助系統可協助駕駛者感知行車周遭環境，進而提升道路行車安全與駕駛環境便利性。本文所提出的解決方案包含硬體平台設計、底層軟體開發、多功能演算法開發與軟硬體系統整合，本平台可支援多種異質感測器、八通道的標準解析度類比影像輸入、三通道全高清解析度數位影像輸入、數位和類比影像輸入與兩組控制器區域網路等功能，並且於符合車用規格標準的R-Car M2系統晶片，成功開發整合基於影像感測器之車道偏離警告系統、前方碰撞預警系統、盲點警告系統與四通道行車紀錄器的功能。本文所開發高度整合多功能智慧行車輔助系統，就現階段測試結果顯示，在各種道路、天候條件的測試結果顯示，平均正確率可達90%以上，在720x480的影像解析度下並且同時啟動四種功能條件下，其執行速度每秒可達25張畫面，具高偵測正確率、高穩定性與低記憶體需求的特性，適合車載嵌入式的應用，此系統持續實車測試驗證中，以利系統可適應更多元的場域。
Abstract
Intelligent driver assistance systems could help drivers perceive the surrounding environment that will enhance driving safety and convenience. The proposed solution includes hardware platform design, board support package development, algorithm development, and system integration. Besides, the platform could support multiple heterogeneous sensors, eight channels of standard definition (Standard Definition) video analogy input, three-channel full HD resolution (Full-HD) digital video input, digital and analog video input, two sets of controller area network (CAN), and etc. With R-Car M2 vehicle regular SoC, the lane departure warning system (LDWS), forward collision warning system (FCWS), blind-spot warning system (BSWS) with four-channel driving record (4Chs DVR) functions are successfully developed. Under various roads and weather conditions, verification result shows the highly integrated system could achieve 90% accuracy rate in the average. As simultaneously enabling the four functions, system performance achieve 25fps (Frame Per Second) with 720x480 resolution videos. This system advantages high detection accuracy, high stability characteristics and low memory requirements that is suitable for automotive embedded application. And, we are continuously field try to adapt more and more condition.
關鍵詞(Key Words)
車道偏移警示系統(Lane Departure Warming System；LDWS)
前方碰撞警示系統(Forward Collision Warming System；FCWS)
盲點偵測警示系統(Blind Spot Warming System；BSWS)
四通道行車紀錄器(Four Channels Digital Video Recorder；4Chs DVR)
相關檔案: 多功能智慧行車輔助系統設計(全文)
