Path Planning for a Connectivity Seeking Robot

We consider the scenario where an unmanned vehicle needs to get connected to a remote station (or another robot). More specifically, we consider the case where an unmanned vehicle is not connected in its current location and needs to incur motion energy to find a connected spot. We are then interested in designing robot paths that are energy efficient (minimum traveled distance) and can result in guaranteed connectivity in realistic channel environments that can experience multipath, shadowing, and path loss. In this paper, we show how this problem can be optimally and efficiently solved, under mild conditions on the paths, using tools from stochastic dynamic programming. Our extensive simulations, with real channel parameters, then confirm that our approach can significantly reduce the traveled distance to connectivity, thus minimizing the total energy consumption.