Applications in Traffic Analysis from Automatically Extracted Road User Interactions with Roadside LiDAR Trajectories

Abstract

Traffic trajectory data extracted from sensors such as LiDAR and camera has birthed new research areas in the field of transportation. Traffic trajectories is defined at the ability to track road users through time and space at high frequency, typically every tenth of a second. Given this higher granularity of micro traffic data, the behavior of road users can be explored to understand the operational and safety performances on the roads. Therefore, researchers have been stretching the limits to what this type of data can be applied to. Most studies look into the interactions of road users as a surrogate safety measure (SSM) to identify potentially dangerous situation using measures such as post encroachment time (PET) or time to collision (TTC); however, not much applications have been explored outside this. This paper seeks to stretch the imagination for what this type of data can be used for using trajectories generated from roadside LiDAR cloud point data. The first application presented in this paper introduces the first automated method to extract headways and determine capacities at roundabouts entry legs. The method provided accurate capacity results when compared to other standard methods. The second application proposes an automated method to extract pedestrian-vehicle yield rates at uncontrolled crosswalks Pedestrian-vehicle interaction (PVI) analyses from trajectory data has been studied using the SSMs mentioned, but an analysis on yield rates using trajectory data has seldom been performed. This research in traffic trajectory applications paves the way for further applications in traffic safety, operations, and planning.