Real time person detection and tracking by mobile robots using RGB-D images

Abstract

Detecting and tracking humans are key problems for human-robot interaction. In this paper we present an algorithm for mobile robots to detect and track people reliably, even when humans go through different illumination conditions, often change in a wide variety of poses, and are frequently occluded. We have improved the performance of face and upper body detection to quickly find people in each frame. This combination enhances the efficiency of human detection in dealing with partial occlusions and changes in human poses. To cope with the higher challenges of complex changes of human poses and occlusions, we at the same time combine a fast compressive tracker with a Kalman filter to track the detected humans. Experimental results on a challenging database show that our method achieves high performance and can run in real time on mobile robots.