Moving Horizon Estimation for 3D Motion Tracking
Tracking the exact 3D movement of a flying and deforming tethered wing is of crucial importance for stable and efficient control. The objective of this PhD project is to design and implement novel estimation methods using the principle of moving horizon estimation (MHE) for real-time applications.
Airborne Wind Energy systems are as many other future relevant technologies complex autonomous systems with a high degree of freedom. One key component to guarantee a robust and fail-safe operation of these systems are precise state estimation algorithms.
Tracking of 3D movements, particularly the movements of tethered airborne objects used in Airborne Wind Energy (AWE) is one of the key challenges in this PhD project. Given that the possibly flexible airborne platform is moving with fast dynamics under varying winds, a major hurdle lies in tracking the exact 3D movement needed for a stable and efficient control.
For this task advanced state estimation algorithms like Moving Horizon Estimation are designed, investigated, analyzed and compared. The input for these algorithms is inertial measurement data acquired at a high frequency by Inertial Measurement Units using state of the art MEMS technology combined with aiding sensor information like GPS. Combining the information from several sensors in one MHE based sensor fusion framework will allow for a precise state and uncertainty estimate at every time instant.
It is obvious that this project involves also research topics like sensor calibration for improved measurement quality, sensor and setup selection for maximizing the information contained in the measurement data, online optimization methods for solving the constrained optimization problem at every instant, effects of distributed optimization in a sensor network and fast embedded implementations for the application of these algorithms in realtime.
The developed algorithms will hopefully contribute to solve the topic of motion tracking for various complex systems and applications and help AWE systems to robustly produce clean energy even under challenging conditions.
Your starting point will be AWE applications and will include the non-linear state and parameter estimation problems as well as selecting a good sensor setup. Your algorithms will be targeted at providing accurate, real-time and robust performance at high output frequencies while running on embedded processors as well as handling of large differential equation systems arising from accurate, high order models or the integration of many different sensors on a non-rigid platform. The results will be demonstrated on an industrial-sized prototype within the AWESCO training network.
Fabian Girrbach, born in 1988 in Calw, Germany studied Mechatronics at the University of Applied Sciences in Karlsruhe. After two years he joined a 1-year Erasmus+ program between the University of Karlsruhe and the Polytechnic University of Valencia, Spain. After finishing his bachelor's degree, he continued his master studies in Embedded System Engineering at the University of Freiburg specializing in robotics and sensors. During his studies he had the chance to work on many different challenging autonomous systems and in particular problems of robot localization and multi-object tracking were subject of his final project works. Now Fabian Girrbach is an industrial PhD student on the Marie Curie AWESCO project at Xsens Technologies in the Netherlands being affiliated with the University of Freiburg, Germany.
Dr. Jeroen D. Hol is senior research engineer. Having graduated with an MSc degree (cum laude) in Mechanical Engineering from the University of Twente, The Netherlands, he received in 2011 a PhD degree in Automatic Control from Linköping University, Sweden. His main research topics are sensor fusion, calibration and machine learning and their applications to inertial sensors and motion capture.