Cooperative Aerial Search and Localization Using Lissajous Patterns

This paper presents a cooperative aerial search-and-localization framework for applications where knowledge about the target of concern is minimal. The proposed framework leverages the sweeping oscillatory properties of Lissajous curves to improve an agent's chances of encountering a target. To accurately estimate the states of cooperative search drones, a discrete-time linear Lissajous motion model approximation is presented in such a way that uncertainty in physical model parameters can be accounted for. These uncertainties are propagated through estimation formulas to improve both agent and target localization relative to a static base station. Numerous experiments conducted in a physics-driven simulation environment show that Lissajous search patterns are a logical and effective substitute for many existing search pattern standards. Furthermore, parametric Monte Carlo simulation studies validate the proposed estimation framework as a more accurate target localizer than other traditional methods which do not account for inaccuracy in the motion model. These techniques hold promise for both static and dynamic target search-and-localization scenarios, allowing for robust estimation by eliminating the need for knowledge of low-level control input to search agents.