智慧城市中視野受阻轉角的合作式碰撞警告

Abstract

近年來，越來越多的汽車製造商推出了配備不同最先進安全系統的車輛。 特別 是，高級駕駛員輔助系統 (ADAS)，它收集車輛周圍的環境因素並幫助避免危險。該系 統會通知駕駛員危險情況，甚至在必要時控制車輛。目前的安全輔助系統大多通過車載 傳感器收集周邊信息。 但單憑這些傳感器可能無法有效檢測十字路口的視線受阻角落。 系統可能無法檢測到車輛從視線受阻的角落接近的潛在危險。本論文描述了一個原型系統，用於提高視線受阻角落的安全性。該系統有兩種類 型的組件:路側單元和多個便攜式接收單元。便攜式接收單元可以安裝在不同類型的車 輛上。該系統結合了三個任務:信息收集、數據分析和數據傳輸。路邊單元將有關被遮 擋角落的信息廣播到便攜式接收單元。該系統將警告道路使用者即將發生的危險，道路 使用者可以採取措施避免危險。本文也提出了一個分析碰撞的方法，路口佔用，用於增 加判斷為碰撞的可能性，更積極地發出警告。同時也設定了一個車輛佔用路口的時間上 限，以減少過多的假警報。先前的研究定義了一個安全性指標，Time-To-Collision (TTC)， 指即將發生的碰撞的剩餘時間。先前的研究建議，TTC的值設置為四秒時，警告駕駛員 是合適的。本系統使用此結果來選擇警告道路使用者的時間。最後，本研究還提出了提 高即時性能的方法。本論文提出的系統也可以部署在沒有 ADAS 的車輛上，以更全面 地提高道路安全性。In recent years, more and more vehicle manufacturers have introduced vehicles equipped with different state-of-the-art safety systems. In particular, Advanced Driver Assistance Systems (ADAS) collect environmental factors around the vehicle and help avoid dangers. The system notifies the driver of the danger and takes control of the vehicles when necessary. Most of the current safety assistance systems collect surrounding information through on-vehicle sensors, but those sensors alone may not be able to effectively detect the blocked-view corners at the intersection. The system may fail to detect the potential danger of vehicles approaching from blocked-view corners.This thesis describes a prototype system design and implementation to improve road safety at blocked-view corners. The system has two types of components: a roadside unit and multiple portable receiving units. The portable receiving units can be installed on different types of vehicles. The system combines three tasks: information collection, data analysis, and data transmission. The roadside unit will broadcast information about the blocked-view corners to portable receiving units. The system will warn the road users of the upcoming danger and the road users can take measures to avoid danger. This thesis also introduces the concept of intersection occupancy, which will expand the coverage of possible collision cases. Previous studies have indicated that it is appropriate to warn the driver when the Time-To-Collision value is set to four seconds, and our system uses this result to choose the most appropriate time to warn road users. In the end, the research work also proposes ways to improve real-time performance. The proposed system can also be deployed on vehicles without ADAS to comprehensively improve road safety.