Navigation of spacecraft around asteroids remains a critical challenge due to communication delays, uncertain surface topography, and the absence of reliable gravitational models. Testing such technologies on the ground using robotic systems enables faster and more reliable development. However, publicly available information on how to create operational frameworks for such testbeds is limited, often due to inter-agency competition or safety concerns at military test sites. This thesis investigates how relative asteroid navigation can be supported and replicated in a ground-based robotic laboratory environment. The work is conducted within the newly inaugurated Guidance, Navigation and Control (GNC) laboratory at TU Delft and focuses on supporting the initial global characterisation phase of an asteroid mission specifically, landmark identification and the generation of circular motion profiles using the laboratory hardware.

A robotic testbed is configured using a 6 degrees-of-freedom (dof) UR16e robotic arm, a 3 dof omnidirectional RB-Kairos rover base, and a 1:10,000,000 scale model of asteroid 433-Eros. A Python software framework is developed to define reference frame transformations between hardware components and generate trajectory points. Control commands for the rover base, robotic arm, and camera are synchronised and executed via ROS nodes. To enhance the laboratory's demonstration capabilities, a Gazebo simulation environment is implemented to test the trajectories. Experimental evaluations assess the open-loop motion tracking accuracy of the system. For a circular trajectory of 1200 mm radius, deviations reached 370 mm (width) and 276 mm (length). Additional uncertainties in the robotic arm’s mounting position introduced mean errors between -39 mm and 66 mm (width), -46 mm and 57 mm (length), and 0.76 mm to 12.96 mm (height). These results indicate that closed-loop control is required to mitigate trajectory errors and improve positioning accuracy.

To support relative navigation, a photogrammetry pipeline is introduced to reconstruct the asteroid model from image data. Surface reconstructions using open-source software yield high-resolution meshes with up to 365,930 faces. Lighting configuration, image coverage, and environmental artefacts are identified as key limitations to reconstruction quality. The results demonstrate that the robotic laboratory can simulate circular arcs orbital motion. To further support relative navigation tasks alignment accuracy and feedback control have to be improved. Recommendations for future work include implementing closed-loop controllers, developing object-detection relative navigation, integrating adaptive lighting, and using the surface reconstruction models to estimate the gravity field of the asteroid model used in the laboratory.

Attitude-determination-and-control faults are one of the major causes of CubeSat failure in education-based missions. One reason is that flight-representative test facilities are rarely affordable. This thesis is a contribution to a low-cost, real-time ADCS test bench that uses educational-licence software and off-the-shelf hardware. A 3U reference CubeSat is modelled in MATLAB/Simulink with pre-existing space-system libraries; the model is auto-coded to C and deployed to Raspberry Pi 5 and Orange Pi 5 Plus single-board computers. EuroSim, installed on each board, schedules and executes the code in hard real time. Software-in-the-loop runs at 100 Hz to validate magnetic-rate detumbling, proportional–derivative Sun-pointing, and an Extended Kalman Filter that fuses magnetometer and Sun-sensor data. Stimulus-file injections confirm the architecture is ready for hardware-in-the-loop expansion. The study details the minimum components, accuracy margins, and build scripts needed to replicate the bench, providing an inexpensive, reproducible path towards flight-representative ADCS testing in educational CubeSat missions.

The traditional approach to space mission design considers deterministic dynamics, relying on a Guidance, Navigation and Control (GNC) system to manage uncertainty dispersions. To simplify the iterative process of trajectory design, reduce the burden on GNC, and increase mission safety, this thesis investigates the use of uncertainty propagation to robustify a deterministically-optimal atmospheric entry trajectory. A hybrid Gaussian Mixture Model-Polynomial Chaos Expansions method is optimised and used to generate analytical expressions of the entry corridor statistics as a function of 16 uncertainty sources and five decision variables. These expressions reduce the design space and directly handle stochastic constraints, simplifying the subsequent optimisation. The method is validated on the HORUS-2B vehicle and reference trajectory, widening the entry corridor margins to 2-sigma bounds with errors in the order of 2%. The output of this thesis is a generic stochastic optimisation methodology that is more efficient than traditional optimisation with Monte Carlo simulations.

This study explores the dynamics of parafoil systems under Titan's unique atmospheric conditions using a dual-framework approach. A 6-degree-of-freedom (6DOF) parafoil-capsule model served as a foundation for verification, enabling the development and analysis of a more detailed 9-degree-of-freedom (9DOF) model. The 6DOF model simplifies the system as a single rigid body, while the 9DOF model captures flexible coupling effects between the parafoil and capsule, modeled as two rigid bodies connected by a spring-damper hinge.

Thorough verification techniques, including energy conservation, angular momentum tests, and sensitivity analyses, validated both linear and nonlinear simulations. A novel quasi-linear eigenvalue analysis provided insights into rotational stability, identifying natural frequencies and damping ratios, while exposing the limitations of linearisation in capturing nonlinear dynamic coupling effects. Sensitivity analyses revealed key dependencies on aerodynamic, structural, and mass parameters. Wind studies evaluated stability under steady and gust scenarios, simulating Titan's zonal wind patterns across various angles and altitudes. Results highlight significant differences between 6DOF and 9DOF models, with the parafoil showing moderated oscillations and the capsule exhibiting amplified responses due to dynamic interactions. These findings emphasize the need to model parafoil and capsule dynamics independently for high-fidelity stability analyses.

The study demonstrates the potential of parafoil systems to enhance stability and precision during planetary descent, reducing reliance on propulsion-based corrections. By systematically comparing 6DOF and 9DOF models, the research underscores the importance of flexible coupling dynamics and detailed environmental modelling for designing robust guidance and control systems for planetary exploration.

Asteroids have been a subject of interest for many years. They hold information on the early stages of the Solar System and planetary formation, they likely contain many rare minerals, and have been closely monitored to detect possible collisions with Earth. Many satellite missions to asteroid have been successfully completed over the years and more are planned for the future. However, a challenge for any asteroid mission is the irregular shape of these bodies. These irregular shapes can lead to highly irregular gravity fields that exert significant perturbations on any spacecraft that has to investigate these asteroids closely. These perturbations make navigation around asteroids challenging. So far, missions relied on human interference during orbiting phases, to correct for the perturbations experienced by the spacecraft. However, there are delays in communication that worsen with increasing distance to the spacecraft. These delays pose a threat as it could take minutes to correct any unexpected perturbations, possibly leading to a catastrophic end of the mission.

To make spacecraft more robust against these perturbations, a new autonomous navigation method was developed to include the gravity field of the asteroid in the position estimate. Several design iterations of the navigation filter have been completed before this study. A Unscented Kalman Filter has been used to estimate the spherical harmonic coefficients of the gravity fields, which was then extended to include a formation of small satellites to estimate the coefficients. A mass concentrated gravity field model was evaluated to estimate the gravity field below the Brillouin sphere, and a convolutional neural network was implemented to automate landmark tracking during navigation. These studies were all completed using numerical models, with limited error sources in the navigation measurements. There have been several studies that found benefits in the use of experimental measurements and integrated tests with flight hardware. Due to limits in numerical models, physical experiments can evaluate robustness in performance better. To further evaluate and develop the performance of the autonomous navigation, an experiment was designed to gather experimental measurements that might be used to further the development of the autonomous navigation filter.

The experiment simulated an orbit around the asteroid 433 Eros. A camera was placed in the position of the satellite to gather navigation images of a 3D printed scale model. The 10:1·10⁶ scale model of the asteroid 433 Eros was created using stereolithography printing. The model was adapted to be hollow, with an internal grid structure for strength. Furthermore, as the model had to be printed in two pieces, the openings were strengthened with stringer sheets to avoid warping. It was found that a significant drying time was needed to fully dry the model before it could be cured and that the model had to be cured at room temperature to keep the dimensions as close to required as possible. The two halves were joined with alignment pins and epoxy putty, after which they were coloured using spray paint. By creating thin layers of spray paint, the surface obtained a rough surface that was more similar to the surface of 433 Eros than the bare plastic.

The 3D printed model was placed in an experimental set-up, that recreated an equatorial orbit around Eros at orbital radii of 30 km, 100 km and 200 km. A stationary camera placed at the spacecraft position at 10:1·10⁶ scale, was used to capture optical navigation images of the surface of the asteroid and the asteroid was rotated to simulate an orbit with a series of image captures. The experiment was lit from above using Sun-like diodes in a dark room to further increase the realism of the experiment. A second set of numerical measurements was created using the 3D modelling software Blender, to act as a benchmark for the experimental measurements created in the test set-up. Several limitations of the experiment were found; the focal length of the camera had to be adjusted which made it an unknown, the distance measurements for the orbits were off by 1 cm, and the exact positions of the asteroid and the camera during the experiment were not recorded for later use. The first limitation could be solved by using a camera with a fixed focal length and the laboratory has since been fitted with a motion capture system which could be used for the latter two problems.

In total 5 datasets were created with 181 measurements each (one image every 2 degrees for a full rotation). The camera had a pixel size of 3.45 μm, a resolution of 1440×1080 pixels and a focal length of 8.6 mm. To evaluate the experimental dataset, the surface landmarks seen in the images were used to estimate the position of the camera. The location of the landmarks was known and the difference in the position estimate between the experimental and numerical datasets was used to evaluate the accuracy of the dataset and thus the experiment. The position estimate was calculated by finding the intersection of the vectors between the camera and the landmarks, solving the linear system with an ordinary least squares. A bias correction was applied to adjust a misalignment in the position of the experiment camera and a sensitivity analysis was performed to find the effect of each component, namely the Cartesian position of the landmarks and the focal length of the camera, on the position estimate. It was found that the position error of the camera in the experimental dataset was in the range of 3000-4600 m for the lower orbits and up to 12000 m for the 200 km orbit radius. It was more than one order of magnitude larger than the numerical dataset, which had an error in the range of 80-450 m for the lower orbits and up to 5000 m for the 200 km orbit. The main causes to this were hypothesized to be uncertainty in the exact camera position, the landmark location with respect to the camera and the focal length. These were all direct results of the experiment set-up. Additionally, the ordinary least squares method used was not sensitive to outliers, which limited the insight into the error contribution of the individual landmarks in the position estimate.

There is still merit in using experimental images over numerical simulations. By completing an experiment, direct sensor inputs could be used in the navigation filter and hardware integration could be tested as well. By using motion capture and robotics to more accurately position all the components of the experiment, the results would become much more reliable. Even if some errors remain inherent in the experiment, it would prove robustness of the navigation filter if it functions under those conditions.

Asteroid missions stand at the forefront of space exploration endeavours. These face a number of challenges, with safe navigation being one of the main issues. The irregular nature of these bodies creates highly perturbed gravity fields that, if modelled incorrectly, can be a danger to the safe navigation of the spacecraft. Therefore, improving the gravity field modelling of asteroids is of key importance.

This research aims to improve the gravity field estimation of asteroids through the use of a satellite constellation consisting of a mothership and a set of CubeSats. In particular, it focuses on the modelling and estimation of the gravity field with spherical harmonics (SH), which makes the research only applicable to the navigation of the spacecraft outside of the Brillouin sphere.

The system design is based on a CubeSat constellation orbiting the asteroid that relays measurements back to a mothership that estimates its state and gravity field through an Unscented Kalman Filter (UKF). The research is centered around 433 Eros asteroid for the model implementation, being an irregular body with known characteristics from the Near-Shoemaker mission.

An end-to-end simulation environment is implemented for the system research. This allows for simulating the orbits of the satellites in a real-world environment considering SH gravity field up to degree and order 15, the perturbation effect of the Sun point mass, and the solar radiation pressure. Additionally, it includes a realistic polyhedral shape of the asteroid including its landmarks. The asteroid model is used together with the modelling of the satellite sensors to estimate the measurements that can be obtained in a realistic scenario including sensor errors, visibility, and communications constraints. Furthermore, the simulation environment contains a UKF filter capable of conducting the gravity-field estimates from the measured states of the constellation satellites.

The research conducts a thorough sensitivity analysis of the scenario evaluating how the constellation design impacts the estimates obtained. This analysis evaluates a number of filter designs, and determines that better performance is achieved when the design models the dynamics of several satellites at the same time. This is followed by an analysis of variance, conducted to obtain a better understanding of the constellation characteristics' effects on the filter estimates.


From the results obtained, a design synthesis is conducted to test the system implemented with an optimised constellation design. Furthermore, the filter design is re-evaluated and improved through the addition of better covariance matrix tuning and the addition of a degree-by-degree estimation procedure that allows to reduce computational load, a main constraint of the system. The results obtained show that the model is capable of accurately estimating the gravity field spherical harmonic coefficients with an error lower than 15% when the filter is tuned properly for SH up to degree and order nine. Additional verification of its applicability has been conducted by testing the system with extreme case asteroids, which show that the system requires a thruster and control system for the satellites to be capable of maintaining stable orbits around the asteroids when these have extremely irregular gravity fields.

Landing a rocket on Earth is a key factor in enabling quicker and more cost-effective access to space. However, it poses significant challenges due to the highly uncertain environment. A robust, reliable, and real-time capable Guidance, Navigation, and Control (GNC) system is essential to guide the vehicle to the landing site while meeting terminal constraints and minimizing fuel consumption.. A 6-Degrees Of Freedom (DOF) flight simulator is developed, including accurate vehicle and environmental models such as variable Mass, Center of Mass and Inertia (MCI), winds, and detailed aerodynamics. Furthermore, not only thrust magnitude and engine deflections are used as controls, but also two sets of aerodynamic fins, as they are present in real rockets. Finally, initial condition dispersions and uncertainties in dynamics, navigation, and controls are included, to assess the robustness of the GNC strategy.
This study applies Meta Reinforcement Learning (meta-RL) to the terminal rocket landing problem. Through repeated interactions between an agent and its environment over multiple episodes, a Neural Network (NN) develops a policy that maps observed states to control actions. Unlike traditional methods, meta-RL leverages both current and past observations to output the optimal action, enhancing the robustness of the policy. Recurrent Neural Networks (RNNs) like Long-Short Term Memory (LSTM), are commonly used in meta-RL for their ability to handle sequential data. However, this thesis investigates also the use of attention-based Gated Transformer XL (GTrXL) networks, which are promising to improve the solution accuracy.
This research confirms this hypothesis, demonstrating that the GTrXL-based policy significantly outperforms the LSTM-based one. This policy is also less sensitive to internal hyperparameters, provided that the NN has a sufficient number of weights and biases to capture all the relevant features. The results indicate that incorporating complex vehicle and environmental models, along with dispersed initial conditions, aerodynamic and wind uncertainties, navigation and control errors, and actuator deflection rate constraints into the training process, yields a robust GTrXL-based policy. This guidance policy is able to meet terminal constraints on position, horizontal velocity, vertical angle, and angular rates in 1000/1000 simulations. The only exception is the vertical velocity which is exceeded on average by only 2-3 m/s. Including a thrust rate constraint mitigates this issue, reducing the average violation to 1 m/s. Finally, developing a terminal patch to increase the thrust magnitude in the last few meters before touchdown completely eliminates the vertical velocity issue, producing a 100% success rate, with all the Monte Carlo runs meeting all the terminal constraints. Moreover, the NN produces a solution in about 6 ms, showing great potential for its real-time use. The results are consistently superior to those of a conventional Guidance and Control (G&C) strategy where a Linear Quadratic Regulator (LQR) controller tracks an optimal trajectory, consuming only 6% more fuel.
Recommendations for future works include improving the reward function to better handle the vertical velocity constraint, and penalizing control effort to have smoother Thrust Vector Control (TVC) and fins control profiles. It is also suggested to expand the simulation scenario, including the unpowered aerodynamic descent.

Due to the increasing interest of the aerospace industry and the scientific community in missions targeting halo orbits, a specific family of periodic solutions within the circular restricted three-body problem, it is highly desirable to find ways to reduce the cost of transfers between these orbits. To achieve this, depending on the mission characteristics, one could opt for reducing the required propellant mass or the time of flight. As such, with two objectives to be minimized, an ideal implementation for a mission designer would provide a Pareto front of feasible transfers instead of the single trajectory commonly obtained with conventional methods. Moreover, the necessary propellant mass can be minimized even further by means of electric low-thrust propulsion due to the significantly larger exhaust velocities. Therefore, the purpose of this research is to develop, implement, and study a suitable approach to obtain a collection of low-thrust transfers between halo orbits optimized in terms of both propellant mass and time of flight.

The approach employs an optimal control indirect method as thrust law which, combined with a heuristic optimizer based on differential evolution, can find a wide variety of trajectories that minimize the aforementioned objectives within the circular restricted three-body problem. Heuristic optimization is employed to remove the dependency of the solution on the provided initial guess and find trajectories in the region of the global minima. To satisfy the demanding boundary conditions characteristic of indirect methods, these constraints are included as a third objective for the optimizer to minimize as well. Due to the tolerance allowed on the constraints, the trajectories are subsequently refined with direct collocation methods. Then, they can be transitioned to a high-fidelity model, taking advantage of the versatility of direct methods. Furthermore, the implementation allows for the inclusion of invariant manifold phases arising from the departure and target orbits to obtain a wider set of Pareto-optimal solutions.

To assess the suitability of the proposed procedure, a specific transfer between two halo orbits around different Lagrange points of the Earth-Moon system was optimized. The results consisted of 100 Pareto-optimal transfers spanning more than 60 days, offering significant mission design freedom. Moreover, the Pareto front outperforms the trajectory found in the literature for a comparable use case by 30\% in all objectives. Next, an optimized trajectory was first successfully verified with the mission analysis software ASTOS and then refined with direct collocation. The refined trajectory exhibited negligible changes in performance, rendering the trajectories obtained with this approach as promising initial guesses for further optimization with direct collocation methods. Moreover, several major perturbations not included in the simplified dynamic system were also corrected with direct collocation. The next steps include: accounting for the eccentricity of the Moon's orbit to fully transition the trajectories to a high-fidelity model, assessing the optimization quality with different use cases, and implementing transfers between different periodic solutions and dynamic systems, such as the Sun-Earth system.

Ensuring the sustainability of future space missions and maintaining continuous access to space necessitates a serious and proactive approach to addressing the space debris dilemma. One crucial measure that can be undertaken is actively protecting against space debris by implementing satellite collision avoidance manoeuvres. This thesis focuses on researching robust methods to perform such manoeuvres, with a particular emphasis on the guidance and control systems of manoeuvrable vehicles. The Starlink constellation is used in this research as a reference scenario, considering its popularity and extensive development, although the developed system is applicable to any satellite. The core of achieving successful manoeuvres lies in the creation of a robust guidance system. This system integrates convex optimisation algorithms with reference tracking mechanisms, enabling the autonomous execution of collision avoidance manoeuvres to be both robust and fuel-efficient. The outcomes of this research represent a significant advancement in space debris mitigation, enhancing space sustainability.

This study analyzes the orbit control performance of solar sails for Sun-Earth sub-L1 halo orbits, which are particularly interesting for solar observation missions. Heliogyros are a promising alternative to the more widely employed Fixed Polygonal Flat (FPF) sails due to their potential to enhance maneuverability. The heliogyro coupled roto-translational dynamics model is tested for the first time in the interest of orbit control, and a new method is developed to validate and verify the model. While achieving a fully functional heliogyro controller remains challenging due to slow attitude control, the developed design shows potential for future improvements. Furthermore, a new Linear Quadratic Regulator (LQR) that employs a hybrid error approach is designed for the FPF sail. Compared to previous research, the new LQR shows improvements in terms of allowable injection errors: the combinations of allowable initial position and velocity errors double with looser performance constraints and almost triplicate for stricter constraints.

Recent years have seen the exponential growth of the number of artificial objects orbiting the Earth. Since space debris can cause substantial damage, measures are investigated to make space operations more sustainable. Amongst these, there is an interest in the development of methods to detect possible collisions between objects in space. Traditional methods require highly accurate propagations with large computational times, so the focus of this thesis is the modelling of faster and more accurate algorithms to enhance collision risk assessment. To avoid long propagations, neural networks have been trained for orbit prediction and uncertainty estimation, with the main goal of calculating the collision probability between two objects. Different analyses have been made to assess the accuracy, stability, applicability and generalisation of the methods developed. The results show much faster collision risk calculations than traditional methods with a similar level of accuracy. It is also demonstrated how a tool which can generalise to satellites with different geometries can be built.

Proximity operations around asteroids are currently at the forefront of space research and industry due to their strategic and scientific relevance. As recent missions increasingly rely on autonomous systems to reduce operational costs, maintaining low-altitude orbits around asteroids without support from the ground represents a key challenge in enabling the transition to a fully autonomous mission architecture.

This research proposes a sliding mode control framework to ensure robustness against navigation uncertainties and gravitational perturbations while accommodating the on/off operating mode of conventional monopropellant thrusters. The designed station-keeping scheme was tested for a range of orbits with varying sizes and orientations around asteroid Eros, demonstrating that the spacecraft can reach altitudes as low as 2 km by relying exclusively on the knowledge of the asteroid’s gravitational parameter. A line-of-sight tracking mode, replacing the conventional two-satellite formation control approach, reduces station-keeping costs and increases coasting time, allowing more opportunities for scientific operations.

Automated control of Earth observation missions does not only allow for a more efficient use of onboard resources but also improved quality on the scientific return and cost reductions in the operations. Current developments in space hardware computational capabilities allow the implementation of onboard optimisation algorithms that can use the state estimations available onboard to dynamically generate control sequences in adaptation to disturbances or unpredictable behaviour.

This study proposes an onboard target management software with the aim of observing a set of targets on Earth’s surface in an automated manner. The development must be carried out within the framework of AAC Hyperion’s integrated attitude determination and control system. The proposed solution is designed to autonomously manage observations and optimise reorientation manoeuvers to designated ground targets, reducing the need for continuous ground intervention and enhancing operational efficiency. The method proposed combines the solving of a Mixed-Integer Linear Programming (MILP) optimisation problem for the scheduling of the observations, while
using successive convex programming to perform optimal control for spacecraft constrained reorientation and target pointing in a real-time manner. These two primary components, together with another task for computing the target visibility windows, operate within a modular framework, facilitating communication with the existing ADCS architecture and making efficient use of onboard computational resources.

The MILP scheduling subproblem selects targets based on priority, visibility constraints, and the required slew times between observations. Initial results identified limitations in MILP performance, particularly as the number of targets or proximity constraints increased, where large computational loads exceeded real-time processing capabilities. Alternative scheduling strategies are suggested for highly complex or densely populated target scenarios. Convex optimisation is found to be a suitable method for the solving of highly constrained manoeuvres in real-time. The reformulation of non-linear systems and complex, combined with the use of successive convexification were used to efficiently solve the spacecraft attitude control problem. This method’s robustness and
capability to detect infeasibility were especially beneficial for the modular onboard software.

The research tested and validated the proposed algorithms using a software-based real-time simulation environment, allowing a preliminary assessment of the observation manager’s performance. On the one hand, the method used for approximating the slew rate was found innaccurate, worsening the performance of the convex optimisation due to its interdependency with the MILP optimisation. On the other hand, results showed that the convex optimization method was effective for achieving precise target orientations, but achieving target accuracy of 0.1° often required refinement through additional onboard controllers, such as a PID controllers, to maintain
target lock during observation. Additionally, the trajectory optimiser demonstrated robustness at an update rate of 0.2 seconds, allowing the spacecraft to realign to targets successfully despite potential deviations in spacecraft state. However, the limited memory and slower processing speed of the ADCS hardware present challenges in extending these methods to onboard use.

Based on the results, the study suggests the modification of the slew approximation, and the use of alternative methods to the MILP, such as dynamic programming, to improve scheduling efficiency in future work. Moreover, testing on the ADCS hardware processor through the use of HIL simulations is essential to faithfully reproduce the onboard performance for analysis. A shift to higher-performance onboard processors or offloading some computational tasks to external processors is recommended to fully support the demands of automated Earth observation operations. This thesis demonstrates the feasibility and potential of integrating advanced optimisation methods onboard Earth observation satellites, providing a foundation for future advancements in autonomous spacecraft operations and contributing to the field of real-time, dynamic scheduling, and control in space systems.

The Jovian system is one of the most attractive destinations for scientific space exploration, with many missions having flown to Jupiter and its moons. Among the latter, Io stands out due to its intense volcanic activity, with plumes extending up to hundreds of kilometers above its surface. Scientists believe that sampling, returning, and analyzing the particles ejected from Io’s volcanoes has the potential to unlock valuable information about Earth’s formation, and the history of the Solar System. For this reason, the concept of an Io Sample Return mission was born at the Jet Propulsion Laboratory, California Institute of Technology, where this thesis was carried out.
The project focuses on the mission design and trajectory optimization of a Discovery-class Io Sample Return concept. It investigates which geometry, maneuvers sequence, and flyby trajectories, can enable the sampling of Io’s Prometheus plume through a single flyby, before returning the material back to Earth.
Firstly, a broad-search of feasible patched-conics round-trip trajectories to Jupiter is conducted, using a simplified two-bodies model. The search incorporates launch, entry, time of flight, mission delta-V, and Io encounter constraints. Two algorithms for the reconstruction of ballistic and targeted flybys are developed and integrated within the trajectories-search workflow. This phase highlights the infeasibility of the 14 years flight time constraint and the 8 km/s maximum Io-relative speed during sampling. The two thresholds are increased to 18 years and 10 km/s, respectively. The solutions-space is progressively filtered with the aid of a primer-vector optimizer, and a single candidate trajectory is chosen for further studies. The solutions from the broad-search are also used to conduct a sensitivity study on alternative plume targets, which highlights Prometheus’ ideal position for sampling missions that avoid Jupiter orbit insertion.
The selected patched-conics candidate is used as initial guess for the numerical propagation and optimization of the end-to-end mission. A high-level trade-off is conducted to select the most suitable approach to propagate the trajectory. Two optimization approaches are then compared. An arc-wise scheme, in which each interplanetary transfer is optimized individually, and an all-arcs method, where the Earth-Jupiter and Jupiter-Earth journeys are each optimized in their entirety. Self-Adaptive Differential Evolution (SADE) and Generational Multi-Objective Evolutionary Algorithm by Decomposition (GMOEA/D) are the two optimizers of choice.
Optimization runs conducted using the patched-conics initial guess prove unable to converge to acceptable solutions. Moreover, single-objective optimization struggles to satisfy position discontinuities requirements, when adopting the all-arcs approach. The patched-conics solution is therefore extended to multi-conic, leading to significant performance improvements. Final results show that GMOEA/D outperforms SADE using the all-arcs approach, and finds a solution that satisfies delta-V, launch C3, entry speed, and position discontinuities constraints. It also improves the total delta-V of the baseline multi-conic solution by about 70 m/s, leaving over 700 m/s of margin on the mission delta-V budget.

IAC will be developing a new Transportable Optical Ground Station (TOGS). It is a telescope that is intended to be used for quantum key distribution in a number of events. Using AO for free-space propagation stabilises the optical signal on a detector significantly, and maximises the received signal by returning near-diffraction limited performance to the optical system and stabilising the channel transmission and noise. The goal of this thesis will be to design a controller for an AO system that will be used for the new Transportable Optical Ground Station (TOGS) of IAC.

The development of reusable launch vehicles in space exploration is a significant paradigm shift in the aerospace industry. Among numerous approaches to fully reusable launchers, a flyback booster, complemented by the emerging importance of spaceplanes, stands out as a viable solution. To outweigh the drawbacks associated with the cost, a reliable mission design has to be performed for the developed flyback booster. The aim of this thesis is to increase the overall understanding of the mission characteristics to decrease the risks associated with fly back trajectories, while optimizing the mission design process. Part of the study concluded that the mission design process can be further optimized by utilizing a detailed design space exploration. It is identified that to achieve such results, it is important to utilize various design space methods as each method reveals a different level of interaction between the factors. Upon understanding the design space better, a trajectory optimization is performed for both powered and unpowered flyback boosters. The findings suggest that while the unpowered capabilities are not sufficient to obtain the desired final distance-to-go, powered vehicle can achieve the mission objectives successfully.

Event cameras are novel bio-inspired sensors that capture motion dynamics with much higher temporal resolution than traditional cameras, since pixels react asynchronously to brightness changes. They are therefore better suited for tasks involving motion such as motion segmentation. However, training event-based networks still represents a difficult challenge, as obtaining ground truth is very expensive and error-prone. In this article, we introduce EV-LayerSegNet, the first self-supervised CNN for event-based motion segmentation. Inspired by a layered representation of the scene dynamics, we show that it is possible to learn affine optical flow and segmentation masks separately, and use them to deblur the input events. The deblurring quality is then measured and used as self-supervised learning loss.

With the alignment of the planets in the 2030s, a perfect opportunity is presented to explore the outer reaches of the Solar System. With this, studying Uranus would become possible. With it being one of the least studied planets in the Solar System, the demand for accurate scientific data is high. This is where Ouranos comes in. This project has the specific purpose to design a scientific mission to Uranus, where measurements will be taken of its atmosphere. The summary below highlights the important parts and results of this report.

This thesis aims to improve the accuracy of re-entry epoch prediction for decaying satellites in orbit. It explores the impact of aerodynamic coefficients, rotational dynamics, and initial state uncertainties on re-entry prediction. By using SPARTA software, the study generates aerodynamic coefficients and considers factors such as object shape, attitude, atmospheric density, and relative velocity. The findings emphasize the importance of precise aerodynamic modelling, particularly the inclusion of lift force for specific shapes like Starlink satellites. Rotational dynamics reduce the range of re-entry epoch predictions by narrowing the average drag coefficient. Initial state uncertainties have a limited effect compared to the average drag coefficient. The research also evaluates the re-entry prediction for a Starlink satellite, aligning closely with existing models after applying a correction factor. Recommendations include sensitivity analyses and investigating different satellite shapes to enhance prediction capabilities. These findings ensure more reliable and accurate re-entry epoch predictions for impact analysis.