應用影像資料分析混合車流下之駕駛行為樣態

Abstract

國內車流環境以混合車流為主，除了小型車、大型車外，尚存在許多機車，不同於小汽車遵循車道規範，機車於路面上行駛的行為差異較大，因此，在機車數量的與時俱增下，導致混合車流中車與車間的交互影響愈加複雜。機車體積小、機動性高之特性，使其可任意穿梭於車道之間，其超車、鑽車等侵略型駕駛行為，為機車之肇事率高、交通紊亂之主因之一。因此，探討汽、機車行為樣態之差異乃重要課題。 綜觀國內外相關文獻，大多以問卷、車載系統、模擬器進行駕駛風格之研究，除缺乏針對機車駕駛人之影響探討，亦無加入汽、機車互動環境下之相關變數，鑒於上述，本研究欲分析混合車流下汽、機車之駕駛行為樣態，因此透過空拍影像資料以車流軌跡擷取軟體擷取車輛於路面上的軌跡，其所含大量汽、機車於路段上的行為，可用於進行駕駛行為樣態之分析與探討。 本研究採用集群分析法進行分析，分群變數之使用係考量車流環境類別(時空佔有率、車流方向亂度)、縱向駕駛行為類別(平均速度、瞬間最大速度、平均加速度、平均減速度)、橫向駕駛行為類別(平均橫向偏移幅度、方向變換次數)、互動駕駛行為類別(尾隨比例)，並考量不同車流情境下之影響。結果顯示，混合車流下的汽、機車可分為三種駕駛行為樣態：積極型、穩健型、保守型。對於機車族群，積極型駕駛人行駛速度較快，較多行駛於禁行機車道；穩健型駕駛人之駕駛行為多介於兩者之間；保守型駕駛人速度較慢；對於汽車族群，積極型駕駛人行駛速度較快，尾隨比例較高；穩健型駕駛人之駕駛行為多介於兩者之間；保守型駕駛人速度較慢。而駕駛行為樣態之分析可應用於開發車流模擬或進行安全分析。

Motorcycles are one of the main transportation modes in Taiwan. In addition, the interaction between vehicles become complex as the number of scooters increased. The characteristics of small-sized and the high mobility enables the scooters weaving in the stream easily. Its act of overtaking, weaving and other aggressive driving behavior is the main reason for the high accident rate and the chaos of traffic stream. Therefore, it is important to discuss the difference of driving behavior pattern for car and scooter. Past research focus on the driving behavior pattern mostly relied on the questionnaire, the vehicle carry system and simulator. They did not consider the influence of rider nor did they include the relevant variables of the interaction environment between cars and scooters. To analyze the driving behavior pattern of cars and scooters under mixed flow, the aerial videography and the vehicle trajectory software were used. The mass amount of vehicles data provides for the driving behavior pattern analysis and discussion. This study focus on using the clustering analysis method. The variable consider traffic environment categories(area occupancy rate and Degree randomness of traffic direction), longitudinal driving behavior categories(average speed, maximum speed, average acceleration and average deceleration), interacted driving behavior categories(tailgating rate) and consider the influence of different traffic flow conditions. It reveals that the vehicles under mixed flow condition can be categorized into 3 types of driving behavior patterns: aggressive, moderate and conservative. For scooter, aggressive driver’s average speed is high and tend to violate the regulation that drive in lane which is not allowed for scooter. Moderate driver’s behavior is in between. The conservative drivers tends to be slow. For car, aggressive driver’s average speed is high. Moderate driver’s behavior is in between. The conservative drivers tends to be slow. The driving behavior pattern analysis can be applied to the development of traffic simulation or safety analysis.