In-Cabin Vital Signs Monitoring System Using ESP Sensors and Machine Learning

Customer: Automotive Industry Manufacturer

Project: In-Cabin Vital Signs Monitoring System Using ESP Sensors and Machine Learning

Product Description

Syntronic partnered with a leading automotive OEM to architect and validate a non-intrusive in-cabin vital signs monitoring system that leverages existing Electronic Stability Program (ESP) sensors combined with advanced embedded machine learning algorithms.

Rather than introducing additional biometric hardware such as radar or camera systems, the solution repurposes validated vehicle sensor infrastructure to detect heart rate, respiratory rate, temperature indicators, and infant distress events in real time. By combining advanced signal processing, motion compensation algorithms, and edge-based ML inference, the system operates within automotive-grade hardware constraints while meeting functional safety and data privacy requirements.

Through cross-domain expertise in automotive electronics, embedded software, AI/ML engineering, and connected vehicle architectures, Syntronic enabled the OEM to expand its ADAS and in-cabin sensing roadmap while maintaining cost efficiency and architectural scalability aligned with Software-Defined Vehicle (SDV) strategies.

Scope

Syntronic led the complete engineering lifecycle from concept validation through prototype deployment and demonstration. The engagement encompassed ESP sensor signal characterization, feasibility studies, embedded C/C++ development, real-time signal processing pipeline design, and system-level architecture definition.

A central innovation of the project was extracting physiological signals from ESP sensor data streams originally designed for vehicle dynamics control. Syntronic engineers conducted detailed sampling rate optimization, frequency-domain analysis, signal filtering, and noise isolation studies. Proprietary motion compensation algorithms were developed to separate micro-vibration signatures associated with heartbeats and respiration from macro vehicle motion under real-world driving conditions.

Machine learning models were trained on large datasets collected across diverse driving scenarios, accounting for acceleration and deceleration events, road surface variability, temperature fluctuations, and varied occupant profiles. The ML pipeline included feature extraction from filtered ESP signals, supervised training and validation, adaptive calibration across vehicle platforms, and model optimization for embedded inference. Specialized infant monitoring models increased sensitivity and detection confidence in child safety scenarios.

Inference was executed directly on automotive-grade ECUs, achieving sub-500 ms latency to ensure real-time alerts without compromising core vehicle control systems. The system architecture incorporated 5G connectivity for telemetry, cloud-based data ingestion for model validation, and alignment with automotive cybersecurity, functional safety considerations (including ISO 26262 principles), and data privacy frameworks such as GDPR.

By reusing existing automotive hardware and validated vehicle networks (e.g., CAN/LIN/Ethernet architectures as applicable), the solution minimized homologation complexity and reduced overall BOM impact while preserving system integrity.

Results

The collaboration resulted in a validated prototype demonstrating high-accuracy physiological detection under dynamic driving conditions. The system achieved 95% accuracy in heart rate detection and 90% accuracy in identifying infant distress, with real-time processing latency below 500 milliseconds.

By leveraging existing ESP sensors instead of adding new hardware, Syntronic reduced development costs by 40% and shortened the prototype timeline by 30%, accelerating the transition from concept to demonstrator. The hardware reuse strategy lowered architectural complexity while maintaining compliance readiness for automotive safety and data privacy standards.

Strategically, the project established a scalable technical foundation for in-cabin health and wellness features within the OEM’s broader ADAS and connected vehicle roadmap. The customer gained a differentiated, AI-enabled occupant safety capability while maintaining alignment with cost, regulatory, and system integration constraints — positioning them at the forefront of intelligent automotive safety innovation.

Performance Metrics

• High accuracy in heart rate detection
• High accuracy in infant distress detection
• Low real-time processing latency
• 40% reduction in development costs through hardware reuse
• 30% reduction in prototype development cycle