混合自動駕駛環境中 V2X 速度建議系統之成效評估：人本導向之研究

Abstract

高速公路的交通流經常因「幽靈塞車 (Phantom Jams)」而陷入不穩定，此現象主要源於人類駕駛的反應延遲。儘管聯網自駕車 (CAVs) 被廣泛視為解決此問題的終極方案，但受限於混合自動駕駛環境的複雜性與法律規範，其全面部署仍難以在短期內實現。為解決此過渡期缺口，本研究提出一套速度建議系統 (SAS)，透過 V2X 通訊技術賦予人類駕駛車輛 (HDVs) 前瞻性的預判能力。 鑑於現有標準模型無法重現導致交通震盪的特定人類駕駛失誤，本研究首先在 SUMO 模擬環境中開發了一套「以人為本跟車模型 (HCCFM)」。該架構明確納入了人類駕駛固有的生理與心理限制——具體整合了韋伯定律 (Weber's Law) 與非對稱風險感知——以精確重現真實的駕駛動態。 實驗結果證實，SAS 發揮了強大的「消波 (Wave-Breaker)」作用，將震盪波傳播距離減少了 94.9%，並將總延遲時間降低了 90.4%。在巨觀層面上，該系統在瓶頸場景中使道路容量提升了 21.2%。敏感度分析顯示，效率顯著提升的關鍵規模 (Critical Mass) 位於 50% 的市場滲透率。關鍵的是，穩健性測試證實即便在 50% 封包遺失率下，系統運作效率仍優於純人類駕駛；且其失效安全機制確保了即使在 100% 通訊中斷的情況下，仍能維持零碰撞紀錄。

Highway traffic is frequently destabilized by "phantom jams'' driven by human reaction latency. While Connected and Autonomous Vehicles (CAVs) are widely proposed as the ultimate solution, their immediate deployment is constrained by the complexities of mixed autonomy and legal regulations. To address this gap, we propose a Speed Advisory System (SAS) that empowers Human-Driven Vehicles (HDVs) with V2X-enabled foresight. Since standard models fail to replicate the specific human errors that propagate these instabilities, we first develop a Human-Centric Car-Following Model (HCCFM) within SUMO. This framework explicitly accounts for the innate physiological and psychological constraints of human drivers---specifically integrating Weber's Law and asymmetric risk perception---to accurately reproduce realistic driving dynamics. Experimental results demonstrate that the SAS acts as a strong "wave-breaker,'' reducing shockwave propagation distance by 94.9% and total delay by 90.4%. Macroscopically, the system increases the capacity by 21.2% in the bottleneck scenarios. Sensitivity analysis reveals a critical mass at 50% penetration for efficiency gains. In particular, robustness tests confirm that the system remains more efficient than human driving even under 50% packet loss, and the fail-safe mechanism ensures zero collisions even under 100% communication failure.