Figure 14.

The road user detection and tracking in the IRSU and CV in the Abercrombie St experiment. (a,b) show the image frames from Cam #1 and #2, respectively, overlaid with pedestrians and vehicles bounding boxes and projected lidar point cloud in the IRSU. The detection results from the dual cameras and the lidar were then fused within the IRSU for tracking in 3D space, as visualised in (c). The detection results were encoded into CPMs and transmitted to the CV, where they were decoded, transformed, and used in the CV’s local frame. In the particular traffic scene shown in (c), the generated CPM contains one Sensor Information Container and seven Perceived Object Containers, two for pedestrians and five for vehicles. (d) illustrates the tracking results of the same group of road users within the CV at the same time. Due to the change of visualisation colours in IRSU and CV, the vehicles in orange pillars in (d) correspond to red ones in (c), while the pedestrians in magenta pillars correspond to green ones in (c).