Internship/ Thesis (m/w/d) Advanced methods for spacecraft attitude determination

e:fs TechHub, specifically e:space, uses the outstanding know-how and long-term experience in technology development from automotive in space applications.

The objective of this thesis is to investigate state-of-the-art methods and algorithms from the literature for accurate and efficient attitude estimation of a spacecraft in orbit using usual sensors such as gyros or star tracker. In most of the cases well-established model-based filtering techniques such as Kalman Filters allow to achieve the required performance, particularly in steady state conditions. However, recent research on learning-based and adaptive estimation techniques have shown some potential on the state estimation accuracy when used to learn unknown dynamics, including sensors error properties. For instance, new approaches use Machine Learning (ML) for sensor calibration or augmentation of Kalman Filters. An ADCS (Attitude determination and control system) test bed and a satellite prototype has been already developed and used at e:fs. An important part of this work will also be the further development, testing and expansion of the existing testbed.

Within this thesis, existing methods and their ML-based extensions will be investigated. One or two most promising algorithms will be implemented on the ADCS test bed as well as using a simulation environment for attitude determination considering realistic spacecraft dynamics and sensor behaviour. Results will be compared to those using classical algorithms and conclusions will be drawn.

• Literature survey on modern methods of advanced attitude determination

• Theoretical analysis and adjustment of available algorithms for the relevant scenarios as well as derivation of algorithms extensions for application on a spacecraft. The algorithms will consider realistic disturbances and unknowns in a spacecraft

• Implementation and validation of the algorithms using a simulation

• Further development and expansion of an existing ADCS test bed and a satellite prototype

• Implementation and validation of the algorithms using on the satellite prototype

• Documentation, conclusion, and presentation of the results

• Ongoing studies in aerospace, mechatronics, electrical engineering, mathematics or computer science and an interest and enthusiasm for research-oriented development in space

• You also have very good skills in estimation and control theory, optimisation, sensor data fusion

• You have a well-structured and results-oriented way of working

• Your programming skills in Python or/ and Matlab are very good and your experience with ROS and C++ is a plus

• Your knowledge in Machine Learning and your experience with implementation of neural networks is a plus

• Very good English or German skills are a must have