Optimized Mobile Indoor Robot Navigation through Probabilistic Tracking of People in a Wireless Sensor Network

Abstract:
In this work it will be shown how the navigation capabilities of a typical mobile indoor robot can be significantly improved by information gathered through a smart environment. Concretely, people moving in this environment are sensed using a network of very simple and cheap wireless sensor nodes and then tracked using Monte Carlo filtering techniques. The maximum permissible velocity of the mobile robot is adapted dynamically according to the current position of the people in the environment. If the path the robot is to take is clear the velocity will be increased, resulting in significantly shorter traveling times. The system is designed to be robust and to also operate when single nodes of the network or even the network at a whole fails. The approach has been evaluated in several simulation experiments and has shown to significantly improve the mean transportation speeds of a mobile robot. Keywords: Indoor Navigation, People Tracking, Wireless Sensor Network, Sensor Fusion, Bayesian Filters