Next-Gen ADAS with Confidence-Aware Radar Vital Sign Monitoring in Dynamic Scenarios

Session details:

The session opens with a discussion of why classical radar systems cannot reliably detect vital signs in the dynamic scenarios demanded by next-gen ADAS. A comparative overview of radar-based approaches covers advances in range-bin selection, phase-coherency analysis, beamforming, MIMO diversity, adaptive decomposition, self-calibration, and high-precision frequency estimation. Performance trade-offs across these methods will be highlighted, showing where they succeed and where they fall short of meeting evolving automotive safety requirements, with Euro NCAP’25 as one example. Building on this foundation, a confidence-aware radar framework will be presented that integrates dual-timescale phase estimation, multi-variate confidence scoring, and adaptive beamforming to maintain robustness against posture changes, occlusion, and multipath clutter. Because ADAS depends on multi-sensor fusion at its core, confidence-awareness is essential for gating unreliable outputs and ensuring trustworthy cross-modal decisions. The session concludes with perspectives on hybrid radar–vision methods for reliable in-cabin safety monitoring.