Development and validation of a universal traffic criticality analysis system based on Surrogate Safety Measures

Abstract

Safety on the road is a prime concern for the public and a high priority in traffic research. New developments in Computer Vision, sensors, Data Science and Information Technology overall have allowed the automatization of classic traffic analysis methods and development of new ones. These advancements synergize with others such as electric vehicles and automated driving in the context of newly found interest in strategical developments for our societies, such as the Smart Cities, in the pursuit of a cleaner, safer and more efficient transportation. In this Master’s Thesis, a proposal is done for the analysis of traffic safety using the existing research infrastructure of the AIM Research Intersection, including video surveillance and other sensors for the automatic recognition of vehicles. This proposal takes the form of an Integral Traffic Assessment System (ITAS), that uses several individual Surrogate Safety Measures (SSMs) for risk prediction according to various criteria and a final merged risk evaluation is returned. This system is developed conceptually, as well as its theoretical framework, and then an instance of it is built and tested with an hour of real traffic data. The results are discussed, it is concluded that they are promising, albeit limited, and several upgrades and adjustments are proposed. Finally, a variety of new ideas for future developments and integration on the ITAS are discussed.