針對混合車流路口管理之觸發式排程與轉向策略指派

Abstract

在人口密集的城市地區，交通擁堵對居民來說是一個重大的憂 慮，而且十字路口的事故也是常見的現象。因此，提高道路效率和交 通安全是全球政府的主要關注點。預計「聯網自駕車」(CAVs) 的發 展將解決這些問題。目前已經提出了各種 CAVs 十字路口管理演算 法。此外，一些論文還考慮了人工駕駛車輛 (HVs)。 在混合交通情況下，一種基於預約的排程方法 H-AIM 提出了一 個積極的「綠色軌跡」概念，並表現出顯著的成效。然而，H-AIM 的交通模型以輪流方式調度不同方向的紅綠燈，這在交通稀疏的情況 下會導致性能不佳。因此，我們提出了一個觸發機制，允許十字路口 管理器 (IM) 檢測到 HVs 所在的車道，並相應地觸發該車道的紅綠燈。 我們還提出了一種比較不同轉向策略組合的方法，並評估各種排 程方法在這些組合下的效能。此外，為了鼓勵公眾轉購買自駕車以提升行車效率，我們分析了不同轉向策略組合對 CAVs 的偏好程度。接 著我們將這些分析整合，以確定不同情境下最佳的轉向策略組合。最 後實驗結果顯示，我們所提出的基於觸發的排程方法是優於現有的排 程方法 H-AIM，特別是在車流量少的情況下。

Traffic congestion is a significant concern for cities, and accidents are common at intersections. Therefore, improving road efficiency and traffic safety is a major focus for governments worldwide. The development of Connected and Autonomous Vehicles (CAVs) is expected to address these issues. Various CAVs intersection management algorithms have been proposed, and some papers also consider Human-driven Vehicles (HVs). In a mixed-traffic scenario, a reservation-based scheduling method H-AIM proposes a concept of “active green trajectory” and has a noteworthy performance. However, the traffic model of H-AIM schedules traffic lights for different directions in a rotating manner, which causes inferior performance in sparse traffic scenarios. Therefore, we propose a trigger mechanism to allow an Intersection Manager (IM) to detect lanes where HVs are located and trigger the traffic lights correspondingly to that lane. We also propose a method to compare different turning policy combinations and evaluate the performance of various scheduling methods under these combinations. Additionally, to encourage the public to switch to autonomous vehicles for more efficient traffic flows, we analyze the extent that different turning policy combinations favor CAVs. We then integrate these analyses to determine the optimal "turning policy combination" for different scenarios. Finally, the experimental results show that our proposed trigger-based scheduling method outperforms the existing scheduling method H-AIM, especially in sparse traffic scenarios.