Design and Implementation of a Radar-Camera Fusion System for Human Detection and Its Distance Measurement

Human distance detection is essential for improving safety and efficiency in applications such as autonomous vehicles, surveillance, and human-robot interaction. Integrating radar and camera data through sensor fusion significantly improves detection accuracy and reliability while reducing the inherent limitations of each sensor type. This research presents the design of a radar-camera fusion system that reads and processes data from both sources. Instance segmentation is used to identify objects for camera data, while radar data is projected to pixel coordinates through coordinate transformation, allowing the two data sets to be fused into the same space. This fusion results in more complete object information. The developed system achieved 100% object detection accuracy in the experimental scenario and demonstrated 93.4% distance measurement accuracy, with average longitudinal and latitudinal errors of 37.1 cm and 17.5 cm, respectively.