Multi-Modal Edge Intelligence System for In-Cabin Driver Monitoring and Preventive Interventions

This study presents a robust, four-layered multi-modal edge intelligence framework for real-time driver telemetry monitoring and closed-loop intervention, utilizing a decentralized edge architecture to overcome the bandwidth and latency limitations of centralized systems. By pairing localized computer vision pipelines with galvanically isolated physiological electrocardiogram (ECG) telemetry modules, the framework provides a fast, dependable, and non-intrusive in-cabin safety system. Empirical validation demonstrates that the proposed architecture successfully satisfies strict automotive performance constraints. The edge computing pipeline consistently delivers a steady 32–35 FPS spatial tracking frame rate while activating critical physical intervention mechanisms within an exceptional 44–45 ms threshold. Furthermore, out-of-band cloud synchronization and cellular emergency SOS alerts are successfully executed within sub-second thresholds without introducing processing bottlenecks to the primary local safety loops. Future research trajectories will focus on: • Integrating lightweight deep learning models to enhance facial landmark tracking stability under extreme low-light or night driving conditions. • Expanding vehicle dynamics diagnostics by incorporating native On-Board Diagnostics (OBD-II) standard interface support. • Integrating real-time GPS tracking arrays to deliver precise spatial mapping data payloads within the emergency messaging dispatch ecosystem.

[1] World Health Organization. Global Status Report on Road Safety 2023. Geneva: World Health Organization; 2023. [2] Putilov AA, Donskaya OG. Construction of a scale for subjective assessment of momentary sleepiness. J Sleep Res. 2013;22(3):312–321. [3] Guanglong D, Zhang L, Kang S, Wang X, Teng S, Liu PX. A multi-modal fusion fatigue driving detection method based on heart rate and perclos. IEEE Trans Intell Transp Syst. 2022;23(11):21810–21820. [4] El-Nabi SA, El-Shafai W, El-Rabaie E-SM, Ramadan KF, El-Samie FEA, Mohsen S. Machine learning and deep learning techniques for driver fatigue and drowsiness detection: a review. Multimed Tools Appl. 2024; 83(3):9441–9477. [5] Biying F, Boutros F, Lin C-T, Damer N. A survey on drowsiness detection-modern applications and methods. IEEE Intell Transp Syst Mag. 2024;19:7279–7300. [6] El-Nabi SA, Ramadan KF, El-Rabaie E-SM, Emam A, El-Shafai W. A real-time design and implementation of intelligent drowsiness and fatigue recognition system for enhancing driver safety. Eng Appl Artif Intell. 2025b;162:112665. [7] Li N, Zhou R, Krishna B, Pradhan A, Lee H, He J, Jiang N. Non-invasive techniques for muscle fatigue monitoring: a comprehensive survey. ACM Comput Surv. 2024; 56(9):1–40. [8] M. K. Hussein, T. M. Salman, A. H. Miry, and M. A. Subhi. ‘Driver drowsiness detection techniques: A survey. Proc. 1st Babylon Int. Conf. Inf. Technol. Sci. (BICITS). 2021;45–51. [9] H. Iwamoto, K. Hori, K. Fujiwara, and M. Kano. Real-driving-implementable drowsy driving detection method using heart rate variability based on long short-term memory and autoencoder. IFAC-PapersOnLine. 2021. 54;15:526–531. [10] Lamaazi H, Alqassab A, Fadul RA, Mizouni R. Smart edge-based driver drowsiness detection in mobile crowdsourcing. IEEE Access 2023;11:21863–21872.