Evaluation of Traffic Conflicts in Microscopic Simulation Model Using Video Anaytics / Muhammad Nabil Faiz bin Mohd Naim

Thesis submitted in fulfillment of the requirements for the Degree of Master of Science

Abstract
Roundabouts are a common form of road junction and safety analysis is a component in the design and operation of roundabout. This study seeks to evaluate the extent to which traffic conflicts identified through a calibrated microscopic simulation model correspond to real-world traffic conflicts on a roundabout. To achieve this, unmanned aerial vehicles (UAVs) and video analytics were employed to collect and extract real-world traffic data, including vehicle trajectories on a local roundabout. The collected data were then compared with the outputs of a calibrated microsimulation model, with traffic conflicts analysed using the Surrogate Safety Assessment Model (SSAM).
The findings indicate that the number of conflicts identified from real-world trajectories is higher than the number of conflicts identified from a calibrated simulation model with a 51.8% error. These differences persisted despite adjustments to key parameters and sensitivity to vehicle position changes in successive video frames. By comparing with results from other researchers, this is thought to be due to the limitation of microsimulation model to replicate actual conflict occurrence mechanisms despite calibration. The difference in conflict frequency may stem from the simulation model's adherence to collision-avoidance principles through car-following models and priority rules, potentially compromising its ability to accurately represent naturalistic road user behaviour.
However, it is also important to note that this study focused solely on critical gap as the input parameter calibrated that uses entry capacity as a performance measure. A review of existing literature suggests that additional calibration parameters could significantly influence the identification and quantification of traffic conflicts. These include queue lengths, SSAM threshold values, safety distance factors, and desired deceleration rates.
Future research could explore the impact of calibrating these parameters to enhance the accuracy and reliability of microsimulation models in capturing real-world traffic conflicts.

Thesis (Master) - Universiti Teknologi Brunei, 2025

Includes bibliographical references from page 82-87