Machine Learning Methods for Radar-Based People Detection and Tracking by Mobile Robots

This paper reports a machine learning approach for people detection and tracking in indoor environments using a compact radar system deployed by a mobile robot. The set-up described in the paper includes a series of experiments carried out in an indoor scenario involving walking people and dummies representative of other moving objects. In these experiments, distinct learning models (a neural network and a random forest) were explored with different combinations of radar features to achieve person versus non-person classification.