Multilayer Modeling of Driver Behavior Based on Hierarchical Mode Segmentation - Digital Signal Processing for In-Vehicle Systems and Safety - page 61

1

0.5

0

1

0

-1

1

0

-1

1

0

-1

1

0

-1

1

0.5

0

1

0.5

0

1

0

-1

1

0

-1

200 300 400 500 600 700 800

Time[sec]

Fig. 4.5 Observed profiles and mode segmentation result (Examinee A, five modes)

displacement from the lead car, if the examinee's car changes the driving lane,
these variables change discontinuously.
The clustering results for the two modes modeling are indicated by colors in
Fig. 4.4 , while the clustering results in the case of five modes modeling are indicated
in Fig. 4.5 . Thus, mode segmentationworks well. In order to investigate the behavioral
meaning of eachmode, profiles of the lateral displacement are enlarged in Figs. 4.6(a) ,
4.7(a) ,and 4.8(a) . In addition, data distributions in range - range rate space - are
shown in Figs. 4.9(a) , 4.10(a) and 4.11(a) . The horizontal axis highlights the range,
while the vertical axis features the range rate. It is evident in these figures that the two
modes of examinees A and B can be understood as “Following on Cruising Lane +
Passing” (Mode 1: FC + P mode) and “Following on Passing Lane + Returning”
(Mode 2: FP + R mode), respectively. This result implies that symbolization of
behavior can be achieved based on the “dissimilarity” of the underlying dynamics.