With the escalating demand for efficient traffic management and the increasing complexity of traffic control, diverse sensor technologies have been implemented to measure traffic in real time. The road-side LiDAR emerges as a novel technology addressing the data gap in multimodal traffic analyses. LiDAR sensing return time to precisely capture distance and reflectivity, generating point cloud data encompassing all traffic trajectory information. It overcomes challenges posed by illumination conditions like light, dust and fog, which often affect camera sensor performance. In addition, LiDAR sensing minimises the effect of changing object position and angles, simplifying object detection and recognition.

This paper tackles the challenges of analysing LiDAR-derived traffic data by proposing a method for traffic trajectory data enrichment. The methodology followed includes creating a semantic map, bridging the physical space and raw data, transforming from a local to a standard Coordinate Reference System (CRS) and enriching data trajectory representation. Three use cases are presented based on the dataset obtained after enrichment: object classification, permissible directions violation detection, and traffic flow density. The proposed method is validated using traffic data from a LiDAR system of 6 sensors located in one of the busiest intersections in Sofia, Bulgaria. The raw sensor data is processed by a fusion box called the Augmented LiDAR Box, delivering time series frames with labelled moving objects in .osef format. The results prove that the proposed data enrichment method successfully transforms the trajectories into semantic sequences, opening up new avenues for trajectory analysis and intersection traffic micro-modelling.