Knowledge Acquisition from Pedestrian Flow Analysis using Sparse Mobile Probe Data

Abstract Autonomous vehicles require high-level semantic maps, which contain the activities of pedestrians and cars, to ensure safe navigation. High-level semantics can be obtained from mobile probe sensor data. Analyzing pedestrian trajectories data is an effective approach avoid collisions between autonomous pedestrians. Such analyses generate new information such as behaviors in violation traffic regulations. However, significantly sparse noisy, making it challenging analyze activity. To address this issue, we propose multiple daily graph-based approaches treat noisy for estimating flow based on improve sparseness data, are fused. After that, a graph created enhance region’s coverage by connecting indicating This proposed successfully trajectory Moreover, was possible identify potential locations where tend cross street analyzing flow. The results indicate that 83% well-known regions corresponded with those extracted using approach. Furthermore, map along 1-km road presented. expected essential understanding different scenarios interactions individuals vehicles.