Improving Robustness of Robotic Networks using Consensus and Wireless Signal Strength*

:When using multiple networked autonomous mobile robots to sense a given area online, it is of high importance to assure the network connectivity even when the robots move in the scope of their application target. In previous work we have used a topological graph property known as bi-connectivity to turn the multi-robot network robust to dropout faults while the robots were sensing a certain area or tracking a target. In this paper we continue that work by including wireless signal strength information as a measure of link quality, adapting the robots' positions to make sure each robot is connected to any other robot by at least two separate routes over time. The proposed algorithm is decentralized and based on consensus theory. We use a solution of the Traveling Salesman Problem to turn the initial arbitrary topology graph into a virtually bi-connected one. We validate our method with simulations that consider asymmetric and intermittent attenuation in the propagation model. The obtained results show that without our method the network can be disrupted while our method compensates unexpected attenuation. This work is particularly suited to networks of UAVs that sense a target area, online.