Automatic Evaluation Based On Traffic Conflict At No Signal Control Intersections Security Risks

As important components in road networks, road intersections are also experiencing a high frequency of road accidents. Usually, it is necessary for traffic engineers to evaluate the safety level of the intersection before and after the implementation of an engineering project which aims at improving the safety level of the target intersection. By doing so, they can draw firm conclusions about the effectiveness of the project. Traffic safety diagnosis has been traditionally undertaken using historical collision data. However, there are well-organized problems of availability and quality associated with collision data. Because of those problems, the observation of traffic conflicts has been advocated as an alternative or complementary approach to analyze road safety from a broader perspective than collision statistics alone. Though the Traffic Conflict Techniques (abbreviated as TCTs) has been introduced into China for more than twenty years, few researchers use automated traffic conflict collection methods to perform road safety evaluation. In order to promote the widespread use of TCTs, this dissertation aims at building an automated road safety analysis method based on TCTs.First of all, by reviewing related literatures, the relationship between traffic conflicts and road accidents are elucidated by the author, the feasibility of using traffic conflicts as surrogate road safety analysis methods are evaluated as well. Based on analyzing the mechanism of traffic conflicts, a refined definition of traffic conflicts is delineated. Also, a theory framework which could be used in analyzing the severity level of traffic conflicts is proposed, considering various affecting factors.Since distances based on Euclidean distance are found too simple to accommodate noisy and partial trajectories, the Longest Common Subsequence Similarity (LCSS), preliminary used for nominal sequences, is adopted to extract typical vehicle motion patterns while passing the function area of road intersections. The intuitive idea of the LCSS is to match two sequences by allowing them to stretch, without rearranging the sequence of the elements, but allowing some elements to be unmatched, thus it is less sensitive than other sequence similarity methods. In order to indentify traffic conflicts from traffic flows, interactions between road users are divided into three types:collisions course, crossing course and non-crossing course. Among these interaction types, interacting vehicles on a collision course is the premises for a conflict to happen. Therefore, a method combining use Time to Collision (TTC) and vehicle speeds is used to determine whether a conflict exists. Probability distribution of predicted trajectories and the correspondent TTC is used to describe the distance between the conflict evolves into a crash. Meanwhile, the severity level of a conflict is depicted considering relevant vehicle speeds. The evaluation model which could be used to describe the severity level of two interacting road users is developed by aggregating crash probability and the severity of the conflicts. Furthermore, a crash risk model used to analysis of the safety level of a whole road intersection automatically is proposed.Finally, a case study is performed by the author to evaluate the practicability and validity of the proposed model. Micro traffic simulation models VISSIM and Matlab, mathematics analysis software, are used to evaluate the safety level of before and after the installation of speed bumps at a T-intersection. The results clearly shows that the proposed model have a great potential for future automated road safety analysis.