透過科技輔助評估人行環境之無障礙性

Abstract

本研究旨在利用科技輔助方法評估人行環境的無障礙性。人行道與騎樓皆是臺灣重要的行人空間，然而許多城市的都市規劃仍然不足。例如，在新北市，一些人行道被劃設為機車停車位，導致行人經常需要在人行道與騎樓之間交替使用。騎樓結合了遮陽、避雨以及提供通行的功能，是步行的良好選擇。然而，其無障礙性經常受到人為佔用或設計不良的影響。 為了解決這些問題，本研究採用運動相機與陀螺儀進行現場數據收集，並運用幾何分析方法估算步道寬度。此外，本研究納入輪椅運動數據（翻滾角、俯仰角、偏航角）進行進一步分析。同時，透過層級分析法（AHP）設計問卷，以量化無障礙性指標的權重，進而建立一套完整的量化評分系統。 研究結果顯示，寬度估算方法的平均誤差為 1.37 %。在評分系統中，通行的連續性被識別為影響無障礙性的最關鍵因素，其次是路面情況（如坑洞、鋪磚縫隙等等）。此外，相較於傳統方法，科技輔助評估方法在數據收集與分析上可節省約80 % 的時間成本。在新北市永和區進行的實驗中，仁愛路、中山路一段和竹林路等路段的無障礙性得分相對較高。 本研究證明了科技輔助評估方法的有效性與效率，為未來都市無障礙空間規劃提供了實務參考。未來建議發展即時影片分析技術來進行寬度估算，並引入深度學習以提升輪椅運動模式識別的精確性與適用性。此外，探索更多物理性與非物理性因素納入評分系統，使其更貼近實際應用需求。

This study aims to utilize technology-assisted methods to evaluate the accessibility of pedestrian environments. Sidewalks and arcades are both essential pedestrian spaces in Taiwan; however, urban planning in many cities remains insufficient. For instance, in New Taipei City, some sidewalks are designated as motorcycle parking spaces, often forcing pedestrians to alternate between sidewalks and arcades. Arcades, which combine the functions of providing shade, shelter from rain, and pedestrian passage, are a practical choice for walking. However, their accessibility is often compromised by encroachment or poor design. To address these issues, this study employs action cameras and gyroscopes for on-site data collection and applies geometric analysis to estimate pathway widths. Wheelchair motion data (Roll, Pitch, Yaw) is incorporated for further analysis. Additionally, questionnaires designed using the Analytic Hierarchy Process (AHP) quantify the weight of accessibility indicators, leading to the development of a comprehensive quantitative scoring system. The results indicate that the width estimation method achieves an average error of 1.37 %. Within the scoring system, continuity of passage was identified as the most critical factor affecting accessibility, followed by surface conditions (such as potholes, paving brick gaps, etc.). Furthermore, the technology-assisted evaluation method saves approximately 80 % of the time required for data collection and analysis compared to traditional methods. Experiments conducted in New Taipei City’s Yonghe District revealed relatively high accessibility scores for sections such as Ren’ai Road, Zhongshan Road Section 1 and Zhulin Road. This study demonstrates the effectiveness and efficiency of technology-assisted evaluation methods, providing practical references for future urban accessibility planning. Future recommendations include developing real-time video analysis techniques for width estimation, incorporating deep learning to enhance the precision and applicability of wheelchair motion pattern recognition and discovering more physical and non-physical factors for scoring system to better suit real-world applications.