search

JRasmusBm / 2019_Pervasive_RoadRage

1 stars 0 forks source link

Paper Four: A Critical Review of Proactive Detection of Driver Stress Levels Based on Multimodal Measurements #9

Closed JRasmusBm closed 5 years ago

JRasmusBm commented 5 years ago
  1. A Critical Review of Proactive Detection of Driver Stress Levels Based on Multimodal Measurements
JRasmusBm commented 5 years ago

Due to the ever-increasing fatality rate and economic losses related to driver stress, an in-vehicle system for early detection of driver stress levels is essential to increase drivers’ safety

It at least validates our interest.

I only skimmed section 4, too much maths :angel:

Chronic stress is a serious issue for professional drivers, as it adversely impacts on their general physical and mental health.

Very interesting. I only thought of the risk of stress because it results in worse driving performance. It of course has huge implications for professional drivers!

This is report is a very good compilation of many other reports.

Therefore, HR and HRV responses toward stressful situations cannot be the same under different driving conditions.

So if we're measuring heart rate or heart rate variability, we need to account for that.

There are different statistical EDA features that have been commonly used to analyze driver stress, such as mean, standard deviation, mean amplitude, peak rise time, peak amplitude, first absolute difference, and mean of the first difference (see able 1).

Lots to think about...

Tonic component or skin conductance level (SCL) can reflect psycho-physical activation and varies between individuals.

Although SCR response is one of the reliable indicators of driver stress, intersubject variability between drivers can lead to fluctuations in the values of SCR features

So it would have to be studied with some ML model. Hope August is up for the task :stuck_out_tongue:

Another common feature in this domain is respiration rate (RSPR). A high correlation between this feature and drivers’ stress levels are reported by Singh et al.

The main limitation of using respiration signals for detecting drivers’ stress levels is the intersubject variability between drivers’ respiration responses. Drivers’ respiration responses to the same stressor can vary based on the size of their chest cavity.

How would you measure that? They talk about putting something up the participant's nose :nauseated_face:

Thus, integration of different modalities to create a more robust and unobtrusive early detection system for driver stress levels, which can reduce current stress-related driving problems, is suggested for future work. Applying effective data collection and analysis methods in order to address methodological issues to build such a system is also recommended for further research.