基於車輛路徑問題模型規劃使用移動式反光量測車輛之道路標線評估流程

Abstract

道路標線提供駕駛者道路環境的資訊，和道路使用者的安全息息相關，因此 必須確保道路標線的功能。移動式反光量測車輛能改善原有的人工量測法，降低 量測路段之作業時間。而有效率的道路標線評估計畫，可以減少系統之檢測成本， 亦有機會擴大檢測範圍，確保更多道路使用者的安全。本研究基於車輛路徑問題 模型，建構可應用於道路標線評估的數學模型，並使用美國佛羅里達州移動式反 光量測車輛計劃的資料進行實證分析，結果顯示道路標線評估模型可提供經過最 佳化的道路標線評估時程表，和歷史資料相比，可用較短的行駛距離及較少的量 測車輛數完成所有任務。本研究同時使用最佳化求解器 Gurobi 和列生成和分支 界定演算法求解，在待測任務數小於 17 個時，列生成和分支界定演算法有較高 的求解效率。

Lane markings provide drivers information of traffic environment, which is critical to the transport safety. Therefore, ensuring the effectiveness of the lane markings is important. The Mobile Retroreflrctivity Unit (MRU) is more efficient than the traditional approach of manual assessment, and thus reduces the operational cost. An efficient schedule of a system-wide lane marking assessment can reduce the cost and may expand the examination area. As a result, there is a chance to enhance the safety of more road users. This study formulates a VRP-based model for lane marking assessment and use data of Florida Department of Transportation’s MRU program as validation analysis. The result of validation displays that lane marking assessment model provides a more efficient schedule compared with historical schedule, that same tasks can be finished in less vehicle amount and travel distance. In this study, both Gurobi solver and column generation and branch and bound algorithm are used to solve lane marking assessment model, and when task amount is less than 17, column generation and branch and bound algorithm has better efficiency when solving the model.