Radar Tracker For Human Legs Based On Geometric And Intensity Features

Abstract

Human leg detection and tracking is a common topic in autonomous driving and mobile robotics. When dealing with heavy rain, fog or smoke, approaches based on vision and laser are denied. Radar sensing can solve problems even in the harshest environments. Many radar tracking methods are based on measuring the human motion state by utilizing the Doppler effect. However, in applications with pivoting sensors or sensors with low bandwidth interfaces, this information may not be available. This paper proposes a method to detect legs with only geometric and intensity-based features and assesses tracking methods to form a leg tracking system to be used in a wide range of applications with special emphasis on emergency response and firefighting.