以MCDM、PLS-SEM與ANFIS探討影響消費者接受自駕車之因素

Abstract

自動駕駛車輛(或稱自駕車、無人車) 藉由最少的人為干預和優化的交通控制系統，具有提高運輸效率和安全性的巨大潛力。伴隨人工智能和實時數據處理技術的進步，推動了實用自動駕駛車輛的發展。自駕車廠商都嘗試去了解影響消費者接受自駕車的潛在因素，部份城市目前正逐步開放無人駕駛電動巴士(AEB)在實際道路上的應用。然而，自動駕駛車輛與消費者的相關研究較少，為了能夠了解這些因素，本研究將利用TAM(科技接受模式)與第二代整合型科技接受模型(UTAUT 2)的集成框架作為研究架構，預測消費者的使用意圖與使用行為。本研究首先將回顧文獻，並邀集專家，利用修正式德爾菲法評估合適的準則與構面，並以決策實驗室(DEMATEL)之網絡分析法(DANP)，定義準則間之影響關係，定義其中最重要的準則與構面，並以偏最小平方法(PLS-SEM)驗證所得之影響關係。其後以適應性類神經模糊推論系統(ANFIS)推衍決策規則，得出對應準則之權重與DANP相互比較之，針對分析結果可對自動駕駛車輛的的發展提出改善方針和建議。Autonomous vehicles (or self-driving cars, driverless cars) that reduce human intervention and optimize traffic control systems have great potential for improving safety and transportation efficiency. Self-driving car manufacturers or system suppliers are trying to understand the potential factors that affect consumers' acceptance of self-driving technology. Some cities are gradually opening up the application of autonomous electric buses (AEB) on actual roads. However, there are few studies on the direct use of autonomous vehicles by users. In order to understand these factors, this study will use the research framework integrated with TAM and UTAUT 2 to predict consumer behaviors and intentions. This study will first review the literature and invite experts to use the modified Delphi method to evaluate appropriate standards and dimensions, and use DANP to define the relationship between standards and define the most important standards and dimensions, and use PLS-SEM to verify the effects obtained relationship. Then, ANFIS is used to derive the decision rule, and the weight result of the corresponding rule is compared with DANP. Based on the analysis results, improvement policies and suggestions can be put forward for the development of autonomous vehicles.