Validating Cyclist Platoon Detection Using Fuzzy C-Means and the DLR-UT Trajectory Dataset

Abstract

This study presents a trajectory-based validation framework for cyclist platoon detection using high-resolution data from the DLR Urban Traffic Dataset. Cyclist trajectories are analyzed to capture arrival patterns at an urban signalized intersection. A fuzzy C-means (FCM) clustering algorithm is applied to identify cyclist platoons based on spatial and temporal proximity. To enable quantitative evaluation, a reference dataset of cyclist platoons is constructed using deterministic geometric and temporal criteria derived from upstream crossing events. The detected clusters are compared with the reference platoons to assess the accuracy and robustness of the clustering approach. The results provide insights into cyclist group formation and highlight the relationship between upstream spatial clustering and platoon formation at the intersection. The study establishes a practical framework for validating clustering-based platoon detection using empirical trajectory data.