Using self-organizing maps to identify turns from driving simulator data
<p>Driving simulators are a main way researchers gather data about on-the-road behavior. However, the quantity of data produced by these simulators forces researchers to rely on algorithms to aid in cleaning and analyzing the data. One example of this is defining whether the vehicle is making...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en |
| Published: |
MSSTATE
2018
|
| Subjects: | |
| Online Access: | http://sun.library.msstate.edu/ETD-db/theses/available/etd-02192018-135211/ |
| Summary: | <p>Driving simulators are a main way researchers gather data about on-the-road behavior. However, the quantity of data produced by these simulators forces researchers to rely on algorithms to aid in cleaning and analyzing the data. One example of this is defining whether the vehicle is making a lane change or turning a corner by broadly categorizing the angle of the steering wheel. A more precise method of identifying these driving maneuvers is described. This method involves using self-organizing maps to consider multiple aspects of user input when determining the existence of a lane change or turn. The results show that while steering angle is the most relevant variable to consider, other variables such as throttle pressure can be used to improve the accuracy of the categorization. This indicates a need for further study into the automatic classification of driving simulator data.</p> |
|---|