Spatial Distribution and Lateral Interaction in Heterogeneous Traffic: A Trajectory-Based Analysis

Abstract

Urban arterials in developing countries accommodate highly heterogeneous, non-lane-based traffic streams whose lateral dynamics remain poorly characterised. This study analyses the spatial distribution and lateral interaction of main vehicle classes on a four-lane undivided arterial in Sri Lanka using high-resolution trajectory data extracted via a YOLOv11 and ByteTrack pipeline with homography-based spatial calibration, achieving sub-metre positional accuracy. Two metrics which proposed are lateral position distributions (Metric A) and pairwise edge-to-edge lateral clearance (Metric B). Results reveal pronounced area-based utilisation with no effective lane discipline, motorcycles and three-wheelers predominantly occupy kerb-adjacent zones while a considerable proportion of motor-cycles encroached upon the central median. Lateral clearance analysis identified motorcycle involved pairs as the highest-risk interactions, with motorcycle-motorcycle encounters recording the smallest mean clearances across all speed categories. These findings provide directly usable inputs for micro-simulation calibration, lane-allocation policy, and safety interventions targeting two-wheeler interactions on heterogeneous traffic arterials.