人工智慧與高精地圖於道路資產管理之實務應用

本文探討人工智慧（Artificial Intelligence, AI）與高精地圖（High Definition Map, HD Map）應用於道路資產管理之實務可行性。訓練資料的質與量攸關AI模型的成效，必須邀請具備道路資產養護與管理領域知識（Domain know-how）者，針對相關數據資料進行標籤（Label），因此十分費時才能累積充足的訓練樣本；另一方面建議採用三維高精度地圖（1/500比例尺）標記各項道路資產之坐標建置其地理空間資訊系統，即利用HD Map的檔案格式儲存道路資產與養護管理資訊，以解決目前道路資產設施圖層坐標系統不一而無法套疊問題。「數位道路」是為「數位國家基礎建設」之首要，透過AI技術與HD Map製圖技術能加速建置以成功推動智慧基礎建設相關應用，例如分析道路服務性能與生命週期，以支援智慧道路之決策系統，讓道路資產設施增加其韌性與延壽、降低維護與管理之成本，以鞏固「國家數位基礎建設」之營運安全性及永續性為目標。

關鍵字

無資料

並列摘要

The feasibility of applying artificial intelligence (AI) and high-definition maps (HD Map) to road asset management practice is discussed in this article. To develop a robust AI model, the quality and quantity of training data are the most critical factors. It is necessary to invite people with domain know-how of road asset maintenance and management to label the relevant data. Thus, the accumulation of sufficient training sample is considerably time-consuming; on the other hand, it is recommended to extract the coordinates of various road assets from 3D HD Map (1/500 scale), and embed them into geospatial information system(GIS); that is, adopting the existing HD Map file format to store the information of road assets and maintenance management. This is to solve non-overlapping issue of various road assets a with inconsistent coordinate systems, ＂Digital Road＂ is of top priority in the establishment of ＂National Digital Infrastructure＂. AI and HD mapping technology can facilitate the process and thus promote related applications of smart infrastructure. For example, to analysis road service performance and life cycle in order to support ＂Smart Road＂ decision-making system. For road assets and facilities, this indicates an advanced resiliency, life-span and reduction in the cost of maintenance and management. Overall, the goal is to secure a safe and sustainable operation environment for ＂National Digital Infrastructure＂.