Passenger Monitoring Using AI-Powered Radar

Low-cost, contactless and privacy-preserving radar-based sensors are gaining attention for in-vehicle passenger monitoring. In this paper, we address two major topics in in-vehicle occupancy detection using radar technologies: 1. Detection of any living body left alone in a car. 2. Passenger counting. For the first one, we propose a novel, easy-to-implement and fast radar signal processing algorithm to detect the presence of alive subjects (e.g., infants, kids, pets), which is 100% accurate in detecting any tiny subject. For the latter one, we address radar low-resolution problems and propose machine learning to be coupled with radar signal processing to count the number of occupants and identify their occupied seats. Our proposed in-vehicle occupancy detection reaches more than 90% accuracy for each seat.