Accurate motion tracking and state estimation of a maneuvering kite are crucial for reliable and robust operation of AWE systems. On the other hand, rapid flight maneuvers are still challenging for state-of-the-art estimation algorithms. The objective of this project is to design, analyze, and implement optimization-based moving horizon estimators. The developed moving horizon estimators for sensor fusion using inertial sensors show a more robust estimation behavior. Moreover, the developed algorithms are particularly suited for online-calibration of the employed sensors and system parameters. The additional computational complexity of the algorithm are a challenge for the real-time implementation of moving horizon estimators on embedded systems with restricted computational resources. These are first essential steps towards a full integration of optimization-based estimators and the promising results will influence the design of the next generation of Xsens motion trackers.