Home My DSpace Submit Browse Search Search Advanced Search Help About DSpace
 Please use this identifier to cite or link to this item: http://hdl.handle.net/1942/21003

 Title: Processing driving simulatordata before statistical analysis by means of interpolation and a simple integral formula Authors: Ariën, CarolineVanroelen, GiovanniBrijs, KrisJongen, EllenCornu, JorisDaniels, StijnBrijs, TomWets, Geert Issue Date: 2015 Citation: Essam Radwan, Essam; Abdel-Aty, Mohamed (Ed.). 2015 Road Safety & Simulation International Conference Proceedings, p. 756-769 Abstract: Driving simulator data can be sampled in function of distance or time. Distance sampling ensures that driving parameters are sampled at a constant distance interval. Time sampling ensures that driving parameters are sampled at a constant time interval. However, importantly, when using time sampling the distance interval is dependent on the driving speed, leading to a negative correlation between speed and number of sampled data points. This paper elaborates more precisely on the (dis)advantages of both sampling approaches as well as on their applicability. In addition, we introduce and illustrate an interpolation technique and a simple integral formula by means of two driving simulator (i.e., STISIM-M400 system) datasets that were collected at our research institute (i.e., the Transportation Research Institute (IMOB) of Hasselt University). In the paper, we argue that the suitability of both sampling approaches is dependent upon the envisaged type of analysis (i.e., point location based analysis vs. zonal-based analysis). For instance, if the interest is in sampling data (for example, mean speed) on a certain road segment (for example a dangerous curve) at a series of specifically located points (for example, at curve entry, curve middle, and curve exit), the nearest sampled parameter value is a very good approximation for the interpolated value. The situation is different in case the objective is to analyze driving parameters in zones of a pre-specified length (for example: mean speed in a zone of 50 m nearby a dangerous curve), rather than at specific point locations. With a time based sampling approach, the biggest concern is the generation of potentially inaccurate mean parameter values since within- and between-subject variations in speed result in a different number of parameter observations for the zone of interest across subjects. To address these issues, we present a piecewise polynomial interpolation technique and a simple integral formula that allows to compute the average for all kind of driving parameters in such a way that a fairer comparison of the behavior of different drivers is possible. URI: http://hdl.handle.net/1942/21003 Link to publication: http://www.ce.ucf.edu/asp/cfp_Rss/Cd/docs/1279431998729.pdf ISBN: 9781495174452 Category: C1 Type: Proceedings Paper Appears in Collections: Research publications

Files in This Item:

Description SizeFormat