Cognitive Computing and Decision-Making of Traffic Intersection Based on Rule Set

Abstract

The decision-making of autonomous vehicles at intersections is of great significance to a safe drive. And it is also popular research content at the moment. In this paper, cognitive computing is integrated into decision-making with the rule-based algorithm to transform environmental information into behavioral results. By establishing the database of traffic signs and rules, the YOLOv5 algorithm is used to recognize traffic signs and combine the rules into rule sets. Based on the Belief Rule Base (BRB) and the Evidential Reasoning (ER) algorithm, the information in the rule set is reasoned and fused. The traffic environment at the intersection is cognitively computed through the rule set. The BRB algorithm assigns weights to each rule and the parameters in the rule which conveniently activated different rules according to the weight calculation. The ER algorithm calculates the belief of each result according to the activated rules. We complete the decision-making of the autonomous vehicle at the intersection through our proposed cognitive model.