Data Fusion

Localization

Localization in GPS denied environments continues to be one of the main challenges towards autonomous navigation of UAVs. Fusing information from multiple sensors to estimate the unknown states has demonstrated to be a powerful tool in order to alleviate this problem, or to improve the accuracy of the available measurements. The preferred methods for data fusion are the Kalman filter (KF) and all its variations (EKF, UKF, etc.), but non-gaussian approaches, such as the Particle filter (PF), may be useful when dealing with some sensors.

Fusing information from multiple sensors using a Kalman Filter allows to improve the estimation, and adds robustness against sensor failures during short periods of time. Computer vision and inertial measurements can be used to estimate the position when GPS signals are not available, or to improve the GPS accuracy.

Publications

[C7] D. Mercado, P. Castillo, R. Castro and R. Lozano. 2-Sliding Mode Trajectory Tracking Control and EKF Estimation for Quadrotors. 19th IFAC World Congress, South Africa, 2014.

[C8] D. Mercado, P. Castillo and R. Lozano. Quadrotors Data Fusion using a Particle Filter. Unmanned Aircraft Systems (ICUAS), International Conference on. USA, 2014.