Improving Intelligent Vehicle Control with a Prediction Model of Passenger Comfort Based on Postural Instability...

Area of Science:

• Automotive Engineering
• Human-Computer Interaction
• Biomechanics

Background:

• Intelligent vehicle performance is increasingly evaluated by passenger comfort.
• Objective measures of comfort are needed to complement subjective feedback.
• Understanding the relationship between vehicle dynamics and passenger perception is crucial.

Purpose of the Study:

• To investigate the correlation between passenger posture dynamics and subjective comfort in intelligent vehicles.
• To identify key kinematic parameters influencing passenger discomfort.
• To develop predictive models for passenger comfort based on posture instability.

Main Methods:

• Conducted a field test with 60 participants across five driving conditions.
• Collected passenger posture, vehicle motion, and subjective comfort data.
• Employed paired sample T-tests and ridge regression for analysis.
• Developed and validated machine learning and deep learning models.

Main Results:

• Higher posture swing speed correlated with increased passenger discomfort.
• Changes in angular velocity were identified as the primary driver of discomfort.
• Machine learning model achieved 87.1% accuracy in predicting comfort.
• Deep learning model achieved 89% accuracy in predicting comfort.

Conclusions:

• Passenger posture instability, particularly angular velocity variations, significantly impacts comfort.
• Intelligent vehicle algorithm design should prioritize mitigating angular velocity changes.
• Predictive models offer a valuable tool for enhancing intelligent vehicle comfort.