Coverage Optimization using Generalized Voronoi Partition

In this paper a generalization of the Voronoi partition is used for optimal deployment of autonomous agents carrying sensors with heterogeneous capabilities, to maximize the sensor coverage. The generalized centroidal Voronoi configuration, in which the agents are located at the centroids of the corresponding generalized Voronoi cells, is shown to be a local optimal configuration. Simulation results are presented to illustrate the presented deployment strategy.

💡 Research Summary

The paper addresses the problem of deploying a fleet of autonomous agents equipped with heterogeneous sensors so that the overall sensing coverage of a given region is maximized. Traditional coverage‑optimization approaches assume identical sensors and rely on standard Euclidean Voronoi partitions, which ignore variations in sensing range, directionality, and reliability. To overcome this limitation, the authors introduce a Generalized Voronoi Partition (GVP) that incorporates sensor‑specific weight functions into the distance metric.

For each agent i, a weighted distance d_i(x)=w_i(x)·‖x−p_i‖² is defined, where p_i denotes the agent’s position and w_i(x) is a positive function representing the local sensing effectiveness of that agent at point x. The generalized Voronoi cell V_i(p) consists of all points whose weighted distance to agent i is not larger than to any other agent. This formulation reduces to the classic Voronoi diagram when all w_i are constant, but otherwise adapts cell boundaries to reflect sensor heterogeneity.

The coverage quality is quantified by an objective functional
J(p)=∫_Ω ρ(x)·