Decentralized Moving Horizon Estimation for a Fleet of UAVs

Current research on Unmanned Aerial Vehicles (UAVs) is focusing on the ability of performing complex tasks by means of cooperation over many aircraft, with the scope of reducing costs and increasing the reliability. However, the use of a cooperative formation deals with several challenges to coordinate a group of autonomous vehicles. A distributed situational awareness becomes an essential requirement towards the objective. In this paper, a Decentralized Moving Horizon Estimator (DMHE) is presented with the scope of distributing the computational burden and limiting the requirements about communication and software complexity besides avoiding the vulnerability of a centralized architecture to faults. The proposed algorithm merges the consensus theory with a moving horizon estimator to overcome Kalman filtering problems in the presence of constraints on any disturbance or state variables. The decentralization of the scheme is obtained by decomposing the overall estimation problem in several optimization sub-models whose convergence is guaranteed by consensus. A preliminary sensitivity analysis was performed to evaluate the results of the proposed strategy and the significance of its main parameters.