In this paper, the personalized and common acceleration distribution characteristics of the human driver are presented by using China-FOT. Firstly, how much data can get a convergent acceleration distribution of human driver is discussed. Multivariate kernel density estimation and kullback-leibler divergence are utilized to compare the acceleration distribution of each driver between datasets which consist of the different amount of data. It is found that 95% drivers can achieve convergent distributions within $200\times 104$ driving data. Secondly, $200\times 104$ driving data of each driver are selected to obtain the acceleration distributions of different drivers. The common acceleration distribution characteristics include: (1) the acceleration distributions of all the drivers follow the so-called dual triangle distribution; (2) the 1st percentile contour interior areas of all the drivers resemble each other and are an approximately diamond shape. The personalized acceleration distribution characteristics include: (1) the 1st percentile contour exterior areas of each driver have many differences, which means that the distribution area and distribution shape differ from driver to driver; (2) the distribution area is strong linearly correlated to the maximum lateral acceleration, and the distribution shape is correlated to the distribution area.
Personalized and Common Acceleration Distribution Characteristic of Human Driver
2018-11-01
840530 byte
Conference paper
Electronic Resource
English
Driver Lane Keeping Characteristic Indices for Personalized Lane Keeping Assistance System
British Library Conference Proceedings | 2017
|Driver Lane Keeping Characteristic Indices for Personalized Lane Keeping Assistance System
SAE Technical Papers | 2017
|