安全是唯一的路

Abstract

世界衛生組織發佈的《2015年全球道路安全現狀報告》表示，雖然行車安全已有所提升，每年仍有約125萬人死於道路交通事故，其中交通事故死者46%為步行者、自行車、兩輪機動車的使用者。有鑑於提昇行車安全的必要性，近年各大車廠已開發許多不同類型的ADAS系統，包含自動煞車、盲點警示偵測、前車防撞以及車道偏移偵測等等。本專題利用異質多核心嵌入式系統，以及多鏡頭的影像資訊，結合多項技術，感知車輛前方的障礙物及距離，成果方面在雙鏡頭測距方面平均誤差為3.1%，亦即精準度為96.9%；而障礙物偵測準確率則可達到93.6%，並提供安全的行車建議動作。

According to the “Global Status Report on Road Safety 2015” published by the World Health Organization (WHO), more than 1.2 million people die due to traffic accidents each year though driving safety has been improved. The report indicates that 46% of the deaths were pedestrians, cyclists and motorcyclists. In light of the need to improve driving safety, a significant number of Advanced Driver Assistance Systems (ADAS) have been developed recently, including adaptive cruise control, blind-spot monitoring, forward collision warning, automatic emergency braking, lane departure warning, etc. This paper used Heterogeneous Multi-core and image data of multi-camera to combine with many technologies. We can detect the different environments and chase objects, then plan the safety route. If there is an emergency situation and the driver can’t do anything, the car will go along the route to avoid obstacles automatically.