CO₂ Dynamics in Taxi Cabins with Driver Behavior Factors

Abstract

We developed an IoT-based in-vehicle CO₂ sensor and measured the cabin environment of three taxis over a 197- day period (July 28, 2021 to February 9, 2022). Analysis of 204,940 data points showed CO₂ concentrations of 799 ppm mean, 678 ppm median, 653 ppm mode, 478 ppm standard deviation, and 5,185 ppm maximum. We confirmed a strong positive correlation between outdoor temperature and CO₂ concentration standard deviation (r=0.91, p<0.01). Significant differences in CO₂ levels were found among 46 drivers (p<0.01), with significant differences detected in 862 of 1,035 pairs (83.3%, Bonferroni-corrected p<0.05). By operation type, transit operations had the lowest CO₂ levels, while during hired operations, levels increased with 1-3 passengers but reversed to decrease with 4 passengers. Correlation analysis between ventilation behavior questionnaires and CO₂ levels showed strong positive correlation with post-break ventilation in winter (r=0.70) and moderate negative correlation with pollen awareness in summer (r=-0.58), but low correlation for other items, confirming a discrepancy between drivers' self-reported behavior and measured values.