Development of a Driving Model That Understands Other Drivers’ Characteristics

Abstract

In this study, a driving model that observes others’ behaviors and that makes its decisions based on the estimated characteristics of the other driver in the situation where two lanes merge on a highway was developed. Additionally, drivers’ probabilities of decisions in relation to their primary characteristics was interpreted. We presumed various drivers have different characteristics such as aggression and caution that affect their making decisions. We simulated the merging behaviors of two drivers in a merging lane and in a main lane after the driver in the merging lane had estimated the characteristics of the driver in the main lane as a typical case. The results of the estimation-success case revealed that two drivers changed lanes immediately after selecting and canceling their decisions several times. However, the results of the estimation-failure case revealed that if a standoff between two drivers occurred, it would take longer to change lanes than in the estimation-success case. Furthermore, the lateral swaying of the cars was worse in the estimation-failure case than in the estimation-success case because the two drivers allocated much cognitive resources and time to monitor of the other car. The importance of understanding others and building a model that understands others in traffic is discussed.