Accuracy and quickness criterion-based driving skill metric for human adaptive mechatronics system

Abstract

Current research is focusing on understanding the driver in order to develop a car driving support system. The car driving support systems must rely on a reliable driving skill algorithm in order to provide optimal support. Previous studies on skill algorithm have combined tracking error and time related variable into algorithm formulation. This method however does not include a car orientation and lateral speed information as an integral part of the algorithm. Two new variables are introduced into the algorithm structure, namely, orientation angle and lateral speed. Nine participants were carefully recruited for a driving test to validate the algorithm. A simulated driving environment was specifically devised for this experiment. A driving track used in this experiment was segmented into five different severities for data analysis. Two fundamental goals have led to the collection of the data and the subsequent data analysis. The first is analysing the variables in relation to the driving task. The second involves data analysis being further extended into analysing the algorithm performance over estimating the driving skill index. The results reveal that the proposed variables are well correlated with the driving task, and improvement in algorithm performance is found to be almost double compared to previous algorithm.