My Master’s Thesis is entitled “People Detection and Tracking from a Small-footprint Mobile Ground Robot using an RGB-D Sensor” and deals with the problem of robustly detecting people on a cluttered environment from an unusual and difficult point of view, where the legs are the most prominent feature.

Our machine learning based, real-time, open-source, ROS-enabled system solves the person detection problem significantly better than alternative approaches, even in challenging scenarios with multiple people in view at once and occluding each other. The system has been deployed on two real robots: Turtlebot, a small wheeled commercial robot supplied to us by the Dalle Molle Institute for Artificial Intelligence (IDSIA - Lugano, Switzerland), and StarlETH, a small quadruped robot developed at the Swiss Federal Institute of Technology Zurich (ETH - Zurich, Switzerland). Preliminary results of this research project have been presented in a video at the “ACM/IEEE International Conference on Human-Robot Interaction 2014 - HRI” and a paper has been presented at the “IEEE/RSJ International Conference on Intelligent Robots and Systems 2014 - IROS” (more info