Camera, radar and lidar sensor synchronization and dataset creation for traffic monitoring

Details

Supervisors

Craeye, Christophe ; Macq, Benoît

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil électricien, à finalité spécialisée

Abstract

Road safety development is still an open field of research. There is a necessity for data acquisition. This work proposes a multimodal sensor for traffic monitoring. The sensor is embedded with a camera, a radar and a lidar. The thesis starts by introducing the roles of the different modalities and continues with the characteristics of all the devices of the system. The whole synchronization principle is explained and optimized to run on a micro-computer. The synchronization uses USB video class protocol, TCP over IP and gptp protocol. After, an example of data processing and combination scheme is proposed. Camera data is processed using YOLOv8 algorithm while radar data is first transformed in Doppler graphs, then filtered and finally processed to extract targets. The lidar data is processed following a custom scheme. Finally the sensor is used for a dataset creation.