Hybrid data fusion for 3D localization under heavy disturbances

This work addresses the problem of stochastic data fusion for systems liable to heavy disturbances, which denote environmental perturbations strong enough to modify the system´s internal structure, including signal interference, sensor faults, physical structure modification, and many other sources of disturbance. In these such cases, traditional filtering methods usually fail to provide reliable estimates because of the highly corrupted sensor measurements. This work proposes to model the data fusion problem for heavily disturbed systems through a hybrid systems modeling framework and presents an online hybrid stochastic filter capable of tracking the system´s state in unfavorable operating conditions. Simulated and experimental results compare the proposed filter´s performance with the traditionally used Extended Kalman Filter (EKF) and show its usefulness as a robust localization filter for an Unmanned Aerial Vehicle (UAV) designed for aerial power lines inspection.