Project description

Roadside LiDAR devices have significantly improved the detectability of multimodal traffic in realtime at intersections. Such high-fidelity observations of traffic and vulnerable road users (VRU) are critical in providing safe and efficient operation of signalized intersections. However, roadside lidars have some drawbacks that limit their application in adaptive traffic control such as massive data volume, data complexity, sensor malfunctions, and occlusion. Improved machine learning algorithms along with sensor fusion capabilities are needed to provide robust detection of traffic and other VRUs. In this research initiation project, CSULB will conduct site selection, sampling, and measurement of the traffic data in different real-world scenarios to assess the prospects for LiDAR to identify VRUs.