Describe sensor fusion architecture for Level 2+ ADAS systems.

Level 2+ ADAS sensor fusion typically includes: Front radar (long-range, 150-250m for ACC), Front camera (object classification, lane detection), Corner radars (blind spot, cross-traffic), Ultrasonic sensors (parking), and optionally LiDAR. Fusion approaches include: Early fusion (raw data level, computationally intensive), Late fusion (object list merging, simpler), and Track-level fusion (Kalman filtering of tracked objects). Architecture must ensure sensor redundancy, handle sensor degradation gracefully, and meet ASIL-B/D functional safety requirements. Real-time processing requires dedicated hardware (Mobileye, NVIDIA) with deterministic latency.