Exploring the fundamentals of using infrastructure-based LiDAR sensors to develop connected intersections

Connected and autonomous vehicle technologies aim to enable vehicles, pedestrians, roads, and infrastructures to communicate with each other and share vital traffic information through networks. However, in the current mixed traffic stage, on-board sensing systems only focus on a limited detection range around the vehicle, thus non-connected road users such as pedestrians who do not have any communication means are passively protected by the systems. In order to develop a human-in-the-loop connected environment at intersections and proactively protect road users who are not included in the current connected systems, the proposed infrastructure-based LiDAR sensing system can be a feasible solution to this problem due to LiDAR sensors’ capability of scanning objects in three-dimensional space and reporting their locations with great accuracy. Since the methods for on-board LiDAR data processing cannot be directly applied to roadside LiDAR data, it is imperative to investigate the essentials of roadside LiDAR ranging from installation strategies to efficient and effective processing point clouds data. In this dissertation, the author first introduced the characteristics of LiDAR sensors and presented the detection range analysis of roadside LiDAR sensors considering the senor’s built-in features and installation techniques. With the appropriate installation of sensors, a systematic approach to extracting pedestrian and vehicle trajectories from roadside LiDAR data was developed in the order of background filtering, object clustering, vehicle/pedestrian classification, and tracking. An application of extracted pedestrian trajectories was demonstrated by predicting pedestrian crossing intention at intersections using trained deterministic and probabilistic prediction models. This pilot study of infrastructure-based LiDAR sensing systems could be a valuable input for various traffic research, which includes cooperation between vehicles and infrastructures for connected/autonomous vehicle systems, vehicle-to-pedestrian crash reduction, smart traffic signals, etc.