ANALYSIS OF TRAFFIC CONFLICTS USING LIDAR TECHNOLOGY

Road safety has traditionally relied on crash data. Crashes are the ultimate and most critical outcome of any kind of road conflict, making them rare and random events. Study of crashes to find the hazardous patterns between road users requires years of data, causing ethical and practical concerns. Alternatively, the observation of interactions and conflicts between road users can provide useful information about the hazardousness and safety of a location. To evaluate the severity of the conflicts, surrogate safety measures are mainly used. Time-to-collision (TTC), being used in this study, is the Time remaining before a collision occurs between two vehicles, assuming that their current trajectory and speed difference remain unchanged. Conflicts are traditionally detected with cameras, while being common, there are still some drawbacks. Since cameras provide 2D images, converting their data into useful 3D trajectories introduces inaccuracies due to perspective distortion, variations in object size depending on distance, affecting the precision and accuracy of measurements. Video monitoring also introduces privacy concerns, since road users can be identified. Additionally, processing video data has high memory consumption and processing time, especially with multiple cameras. Alternatively, LiDAR (Light Detection and Ranging) is a more recent technology that uses laser light to measure distances to objects or surfaces. This is done by emitting laser pulses and measuring the time-of-flight. Acquisition of data using this technology has centimetre-level accuracy, and unlike cameras, no personal data is recognized, and thus no privacy-related concerns exist. Though the LiDAR technology has its advantages, its performance in the field observation and conflict analysis still needs to be studied to ensure its applicability in further implications. In this study, we used LiDAR as the primary conflict analysis tool and GPS (Global Positioning System) as a reference tool for comparison reasons. A series of experiments was designed to evaluate the performance, accuracy, and reliability of a LiDAR-based tracking system. The research began with an assessment of the optimal setup of the sensor to obtain the best of the sensor's visibility range. Subsequent phases focused on validating the system’s ability to construct trajectories using high-precision GPS data as reference. Coordinate system transformations were applied to align LiDAR trajectories with GPS data, making a direct comparison between both sources possible. Finally, the surrogate safety indicator, TTC, calculated by both LiDAR-derived and GPS-based trajectories, was compared to assess the system’s capability in conflict detection. The results from LiDAR and GPS showed an overall consistency, though some interruptions and errors appeared, likely caused by the increased distance from the sensor, which affected the stability of the measurements. It is important to emphasize that the reliability and quality of the collected data largely depend on the survey design of the data collection setup, as the influence of the sensor configuration and experimental design has been clearly observed throughout this study. Overall, the findings suggest that this technology has the potential for conflict analysis; however, its full capability still needs to be further validated under real-world operating conditions.