無線車載安全技術研究

Abstract

在車輛安全應用上，目前市售車輛已應用了大量的影像及雷達等技術，用以輔助駕駛，增加行車安全。但影像及雷達技術皆有受限於直視限制(Line-of-Sight, LOS)的缺點，對於視線外或受阻隔的物體無法有效偵測。相較於現有技術，將無線通訊技術應用於車載安全具有不受直視(LOS)限制的關鍵優勢。在本文中，我們試著將無線通訊技術同時應用於:1.車外行人偵測、2.車內手機偵測。同時以利用無線通訊技術聯結人、車、路，形成無形保護網為目標及願景。 在車外行人偵測上，我們試著打破現有一般人對Wi-Fi不適合用於行駛之車輛、高延遲性、較短傳輸距離之限制。嘗試以Wi-Fi Beacon傳送具極低延遲要求之行人安全訊息。為了證明我們的想法與可行性，我們實作了具此一概念之行人安全app，並實際於多款具代表性之市售手機及Wi-Fi網卡上進行實際路測。結果顯示我們的方法確實可應用於實際行人及車輛環境，並具有600 m高傳輸距離，300 ms低延遲性，以及適用於100 km/h快速移動環境的特點。 在車內手機偵測上，我們利用了幾乎所有智慧型手機都有配備的藍芽技術，並在車內安裝藍芽訊號發射器，同時開發手機APP。該APP可以偵測車內藍芽訊號並判斷手機目前位置，如偵測到手機位於駕駛座時則關閉除了藍芽以外所有之通訊，避免駕駛於開車時因使用手機而有所分心。實際於車內實測結果顯示，經過調降功率後的藍芽發射器在50~60 cm短距離內之訊號強度具線性關係，可用於判斷距離。且因車內駕駛與座椅阻隔，確實可有效達到分辨區隔手機於車內之不同位置。相較於之前利用聲音、GPS、感測器等之相關研究，我們的結果具高正確性，對將藍芽應用於車內定位具初步突破及進展。

There are many technologies which could be used for vehicle safety applications. However, many of them such as radar-based and vision-based technologies, are lim-ited to Line-of-Sight (LOS). It is hard to use them to detect the pedestrians and ob-jects in Non-Line-of-Sight (NLOS) cases. In this thesis, we apply wireless communi-cation technologies which are not limited to LOS to vehicle safety applications. In this thesis, we investigate how to apply wireless communication technologies to two key safety applications: (1) pedestrian collision avoidance, and (2) sensing phone use in-side a car. For the first one, we present a novel Wi-Fi-based Pedestrian Collision Avoidance (PCA) system called WiSafe to protect pedestrians. Instead of using WAVE/DSRC, WiSafe provides a simple, effective, and low-cost solution. Without changing Wi-Fi standards, WiSafe is compatible with 802.11a/b/g/n and can be im-plemented by apps on smartphones or wearable devices. Complete analysis, imple-mentation, and experiments are presented in this thesis. The results show that WiSafe can achieve and even exceed the PCA requirements. For sensing phone use inside a car, we propose a Bluetooth-based system, called BTGuard, to detect and prohibit driver phone use while driving. For a small confined, highly-multipath, and low-loss envi-ronment inside a car, signal attenuation caused by multipath fading and shadow fad-ing may be greater than that by path loss. We first conduct preliminary experiments and model the channel. We then propose Proximity-based Phone Location Detection (PPLD) to detect driver phone use inside a car. A proof-of-concept prototype is also implemented. The experimental results show that BTGuard can have a 95% accuracy and low false positive rate.