A Knowledge-Based Expert System for Pedestrian Safety Improvement at Intersections

Abstract

In response to the rising concerns about intersection safety across the United States, traffic administrators have developed various techniques to create more effective and targeted improvement projects. Among them, Knowledge-Based Expert Systems (KBESs) demonstrate the unique advantage of having low requirements for users' experience and efficient decision-making. Recognizing that existing KBESs often lack comprehensive analysis of the critical factors contributing to pedestrian-involved crashes and the capability to optimize countermeasure selection, this study proposes an enhanced KBES to assist the traffic community in efficiently generating a set of optimal cost-benefit countermeasures to address pedestrian safety risks at intersections. In the proposed KBES, the carefully designed knowledge acquisition process fills two knowledge bases: one containing well-evidenced cause-effect relationships between contributing factors and corresponding Safety Related Intersection Characteristics (SRICs), and the other storing various attributes of a comprehensive list of countermeasures. The first developed inference engine is capable of identifying the contributing factors at an intersection and innovatively quantifying the impact of each of them based on the user input of SRICs. The second inference engine optimizes the countermeasure selection to maximize the expected effectiveness in accurately targeting the impact of those contributing factors while accounting for both budget constraints and users' defined priorities among the countermeasures' attributes. The results of the performance evaluation indicate that the proposed KBES is effective in analyzing contributing factors and recommending countermeasures and can serve as an efficient tool for traffic engineers to develop safety improvement projects at intersections