Multilayer Modeling of Driver Behavior Based on Hierarchical Mode Segmentation - Digital Signal Processing for In-Vehicle Systems and Safety - page 65

Higher Level Decomposition

Following on Passing Lane
+ Returning Mode + Passing Mode

Following on Cruising Lane

Short Range
Following on
Cruising Lane
Mode

Long Range
Following on
Cruising Lane
Mode

Following on
Passing Lane
Mode

Passing
Mode

Returning
Mode

Lower Level Decomposition

Fig. 4.13 Identified hierarchical structure of driving behavior (Examinee C)

input and output signals) and data processing. Since this hierarchy clearly expresses
multiple abstraction levels of human behavior, the proposed framework is expected to
serve as basis for the design of several human-centric systems.

4.4.1 Development of Symbolic Behavior Model
and its Application to Behavior Prediction

In this section, some applications using proposed human behavior model is discussed.
As already mentioned, this hybrid system modeling is considered as a solution to
“symbolic grounding” using symbolization. With this approach, the human behavior
can function as an entity (linguistic source) capable of generating a specific language
(set of symbol strings). This assumption makes it possible to analyze and model the
human behavior along with formal behavioral grammar such as Production Rule.
Furthermore, the grammatical modeling of higher human behavior based on hybrid
system symbolization will make long-term prediction of human behavior possible.

4.5 Conclusions

This chapter has presented a new hierarchical mode segmentation of the observed
driving behavioral data based on multiple levels of abstraction of the underlying
dynamics. By synthesizing the ideas of the feature vector definition revealing the
dynamical characteristics and unsupervised clustering technique, the hierarchical