自行車道路網配置規劃模式：考量公共自行車軌跡資料

Abstract

公共自行車系統在台灣已行之有年，使用量也逐年增加，然而既有的自行車路網可能不足以應付日益增加的需求，因此需要重新規劃自行車路網。過去在自行車路網規劃的研究上，多採用評估、問卷或未考慮實際需求的數學規劃方法來設計，如此無法滿足使用者的實際需求，並在騎乘的過程中失去單車道的保護，有了軌跡資料的幫助就可以填補過去研究仰賴旅運需求假設的研究缺口。 本研究目的在使用資料導向的研究設計來發展多目標規劃模型，利用騎乘者的軌跡紀錄來建立數學規劃模型。軌跡資料來源取自臺北市政府交通局與微笑單車公司合作開啟GPS定位紀錄騎乘軌跡，並利用去識別化的軌跡資料與路網資料疊合來找出使用者對路段的需求，並進一步設計數學規劃模型來求解。本研究目標包含最大化滿足路段使用需求、最大化自行車騎士的使用效益、最小化道路使用者的風險與最小化對非自行車騎士的使用影響 。限制則考量建設預算、車道種類、寬度與變數限制。本研究選擇台北市大安區作為模型的實例分析場域，並針對調控參數進行敏感度分析，在不了解決策者偏好的前提下，使用ϵ-限制法來求解非劣解，作為替選方案提供決策者評選，藉此來研判模型的可應用性與特性，藉此找出合適可供決策者參考的自行車路網配置。 本文是臺灣首次應用微笑單車軌跡資料的單車道路網設計問題研究，主要貢獻為驗證軌跡資料的應用性與實際操作的可行性，並補足過往自行車路網規劃研究對旅運需求仰賴假設的研究缺口，同時考慮不同種類的車道規劃，提供決策者在車道規劃上的彈性。

Public bike system (PBS) uses in Taiwan have been thriving recently. The old bikeway network may not be sufficient for biking travel demand. The bikeway network should be re-designed to meet the travel demand. Researchers in the past mostly applied evaluation-, questionnaire-, or mathematical programming- based approaches without consideration of travel demand to design bikeway networks. Such approaches may not meet bikers’ needs and lead to safety concerns. The trajectory data is helpful to fill the above research gap. This study applied a data-driven approach to develop a multi-objective programming model to meet different stakeholders’ benefits. The trajectory data was collected by the Department of Transportation, Taipei City Government and YouBike Company by opening GPS records. This study utilized the de-identification trajectory data and the roadway network in Taipei City to develop the mathematical model. The proposed model aims to maximize the travel demand covered by bikeways, maximize the utility of cyclists, and minimize the risk and impact on the other road users. Model constraints include budget limits, bikeway types, bikeway width limits, and value ranges of decision variables. A case study in Taipei Da-an District was conducted to verify the model’s effectiveness and applicability, and sensitive analyses of tuning parameters were applied to investigate the influence of policy-related parameters. Without knowing the preferences of decision-makers, this study applied the ϵ-constraint method to find non-dominated solutions. This study is the first research using trajectory data of YouBike 2.0 to develop bikeway networks in Taiwan. The major contributions of this study are verifying the applicability of trajectory data on bikeway network design, filling the research gap of relying on assumptions of travel demands, and further considering the different types of bikeways for decision-makers to make choices.