Marrying Stationary Low-Power Wireless Networks and Mobile Robots in a Hybrid Surveillance System

Stationary low-power wireless networks are able to continuously monitor large areas for long periods of time. If densely deployed, they can accurately detect intruders and localise them. However such density requires high installation and maintenance costs, and no identification of the intruders can be made. On the other side, mobile robots are able to precisely monitor areas and identify intruders when present in their field of view. Advances in multi-robot coordination allow them to cover large areas as patrolling guards. Nevertheless, they usually leave areas uncovered, and they can only work under battery-constrained time conditions. In this paper, we investigate the trade-offs in a practical hybrid surveillance system. A sparse low-power wireless network continuously monitors the environment and roughly estimates events locations, while robots resting on strategic positions wait for warning calls to survey specific areas and accurately observe the ongoing events. As a result, only a moderate accuracy is required from the stationary network, allowing a notable simplification with respect to similar localisation systems. At the same time the robotic team size can be decreased and its coordination simplified. The on-demand activation drastically saves the energy and wear required by the continuous patrolling. In this paper, we provide an experimental analysis of such trade-offs and we report on the lessons learned.