Prediction Accuracy Improvement of Avoidance Driving Behavior Model Based on Game Theory: Introducing Jerk to Refine Acceleration

Understanding avoidance driving behaviors is important for both drivers and traffic managers owing to the high correlation between accidents and human driving behaviors. The most recent research in this area considers acceleration as the implement of decision result, which may cause inaccurate prediction results as acceleration changes continuously. To solve this problem, this study considers the acceleration as a continuously changing variable and utilizes jerk, the change rate of acceleration/deceleration to quantify the change of acceleration. We further applied the jerks in a game theory vehicle avoidance model. This method analyzes driving behaviors in a more detailed magnitude and finally improves the model results. Finally, a case study shows that this work improves the prediction accuracy of the driving behavior model compared to previous research.