Detection and Tracking of Moving Objects Using a Roadside LiDAR System

Abstract

Monitoring vehicles and pedestrians in urban environments, as required in autonomous/assisted driving, smart crossing, and traffic surveillance, can be performed by means of different systems that use camera, RADAR, or LiDAR sensors. The systems based on the use of the most recent high-resolution LiDARs, characterized by high data throughputs, increase the demand for significant hardware resources, in terms of both storage capabilities and processing units. These hardware resources are supposed to be locally connected to the same system, since bulk transfers to remote units for exploiting cloud services would imply excessive latencies. As the computational effort of the processing tasks seems out-of-reach for low-cost processors, powerful workstations are typically deployed to execute them.