UFN: User-Friendly Navigation Framework Based on Systematic Utilization of User-Friendliness Features from Historical Traffic Data

Due to the ever-increasing number of traffic vehicles, navigation services play an important role in daily life. Current studies typically plan the fastest route by two steps: traffic forecasting and route planning. The planned routes may not be safe or comfortable for drivers. Even if some studies consider user-friendliness factors and extract safety and comfort information from historical traffic (HT) data, they simply combine user-friendliness factors with driving time at the beginning of traffic forecasting. It would be more accurate to keep and use all the information until the planned route is obtained. To address this challenge, we propose a user-friendly navigation framework called \textscUFN. \textscUFN extracts user-friendliness features from HT data and systematically predicts all the features without combining them beforehand. Based on the prediction results, \textscUFN plans the most user-friendly route by simplifying and solving a multiple-objective optimization problem. Experimental results demonstrate that \textscUFN is accurate, and the more accurate the predicted user-friendliness features, the more user-friendly the planned route.