Contactless pulse rate assessment: Results and insights for application in driving simulator

Abstract: Remote Photoplethysmography (rPPG) enables continuous and unobtrusive assessment of driver's state, allowing estimation of fatigue, stress, and user experience. Commonly used wearable PPG sensors, while effective, suffer from motion artifacts and user discomfort. This study explores the feasibility of non-contact Pulse Rate (PR) assessment using facial video recordings captured by a Red, Green, and Blue (RGB) camera in a driving simulator. The proposed approach detects subtle skin color variations due to blood flow and compares extracted PR values against reference measurements from a wearable Empatica E4 sensor. We evaluate the impact of Eulerian Video Magnification (EVM) on PR estimation and assess statistical differences in PR between age groups. Data obtained from 80 recordings from 64 healthy subjects covering a PR range of 45-160 bpm are analyzed, and PR extraction accuracy is quantified using metrics, such as Mean Absolute Error (MAE). Results show that EVM slightly improves PR estimation, reducing MAE from 6.48 bpm to 5.04 bpm. Significant differences between PRs in younger and older drivers are observed in both rPPG and reference data. The study also thoroughly discusses bias within Empatica E4. Overall, results support the feasibility of integrating camera-based PR monitoring into driving simulators for real-time subject's assessment.