Autonomous planetary rovers require obstacle detection capabilities to navigate hazardous terrain without Earth-based intervention, yet currently deployed methods are limited. This research develops and validates a complete deep learning system for autonomous rock detection on a resource-constrained planetary rover.
We present a lightweight MobileNetV2-based U-Net architecture with dual attention mechanisms, optimized for edge deployment with only 0.31 million parameters. A new dataset MarsTanYard was created via a semi-automated dataset creation pipeline, enabling efficient annotation of Mars-analogue terrain imagery. Our deep learning network was trained on this dataset and integrated with a ROS2 navigation stack through a modular architecture that transforms segmentation masks into 2D occupancy grids that can be used for rover path planning. Our network achieves a 77% intersection-over-union accuracy for rock segmentation, and physical validation on a testing rover in a Mars analogue environment demonstrates a 94% detection rate for large rocks at close-range. An inference time of 4.49ms was achieved on the target rover hardware using model optimization techniques. The system maintains reliable operation across varying lighting conditions with less than 15% performance degradation.
Results show theoretical collision probability of 7.8 × 10^-7 per rock encounter, enabling months of autonomous operation for typical planetary missions. This work provides an end-to-end validation of the deep learning obstacle detection system, establishing a foundation for enhanced rover autonomy in future Mars exploration missions.

Target-in-the-loop (TIL) systems are key to engaging aerial threats and enabling laser processing. This work simulates a TIL system and evaluates its performance under varying atmospheric turbulence and target distances. The goal is to assess how these factors affect energy delivery and real-world viability. A computational model was developed using standard wavefront propagation to account for turbulence and target surface roughness. The simulation allows controlled changes in turbulence strength and range. An experimental setup was also created to validate the results. Findings show that compensating for surface roughness is crucial, and system performance improves significantly with a controller—boosting peak intensity by up to 60% and power in the bucket by 10%. A tradeoff exists between distance and turbulence strength, but adaptive optics can help extend operational range.

Free-space optical communications on board high-altitude platform stations (HAPS) promises commercial upside for supplying internet connectivity to regions underserved by space constellations. However, platform vibrations and stringent hardware constraints make pointing, acquisition, and tracking (PAT) of optical links difficult to implement. This work investigates the effectiveness of a cascaded control architecture for the combined actuation of a coarse pointing assembly and fine-steering mirror for PAT between HAPS and LEO spacecraft. The laser communications hardware and platform dynamics are simulated in detail, followed by a systematic control design using H∞ optimization. An Extended Kalman Filter combining recorded light measurements from the target and trajectory propagations provide the system with continuous pointing input. Results show that the cascade control architecture can
reject disturbances up to 300Hz in tracking and correct pointing errors up to 40mrad in acquisition. The EKF output accuracy significantly degrades after 15 seconds due to GNSS antenna instability caused by the bending of the HAPS air frame.

Free-space laser communications promise high-data rates for ground-to-space uplinks, but precise beam acquisition remains a critical challenge due to platform attitude errors and atmospheric turbulence. Conventional collimated beacon beams are often constrained by safety regulations, leading to the usage of defocused beams, which also promise efficient acquisition times. This Master thesis conducted at Airbus Netherlands B.V. investigates the near-field turbulence impact of defocused beams on far-field scintillation. The study comprises
theoretical modelling, wavefield propagation simulations, and experimental measurements to examine how varying the transmitter aperture radius affects scintillation relative to collimated beams under different turbulence conditions. The simulations and the theoretical model show that far-field scintillation grows as the beam becomes more defocused with increasing transmitter sizes for different turbulence conditions. Results indicate that while moderate defocusing doesn’t affect propagation performance, excessive defocusing in strong turbulence conditions amplifies irradiance fluctuations, potentially degrading acquisition efficiency from the computed amplitude statistics. Preliminary Experimental breadboard findings confirm the general trend but highlight discrepancies in absolute scintillation values due to laboratory downscaling constraints.

As demand for high-bandwidth data transmission continues to increase, future satellite constellations can leverage the benefits of free space optical communication, which provides significant advantages compared to traditional radio frequency systems. Despite the development of satellite laser communication terminals (LCTs) in recent years, one major challenge remains; coupling the incoming signal into an optical fiber. In addition, most terminals use intensity modulation to encode data, which is limited in its ability to reach high data rates. So-called coherent modulation schemes offer superior performance but are not widely implemented in satellite LCT’s yet. A system that avoids fiber coupling while using coherent modulation could simplify designs, reduce complexity, and improve performance. This thesis presents the design, implementation, and evaluation of such a coherent free space optical detection system. Differential Phase Shift Keying (DPSK) was chosen as the modulation scheme due to its balance of power efficiency and implementation simplicity, delivering performance comparable to other coherent schemes with a less complex receiver architecture. The system was implemented using bulk optics and tested in a laboratory environment. A Delay Line Interferometer (DLI) was used for DPSK demodulation. The system successfully achieved a data rate of 100 [Mbit/s], demonstrating the feasibility of DPSK-based coherent free space detection. Some performance limitations were identified, with DLI instability emerging as the most notable, which is primarily caused by the long length required for this data rate. This instability resulted in intermittent bit error rate values exceeding the required threshold. These findings show that DPSK demodulation using bulk optics is favored toward higher data rates, and thus shorter delay lines. Recommendations for future work would be to operate at a higher data rate, and implement active stabilization for operational systems.

Laser satellite communication effectively addresses high data rates, low latency, high security and license-free operation of down to Earth and inter-satellite data links, crucial for the growing data demands in space amid radio frequency spectrum congestion. However, this approach presents challenges, notably in meeting micro-radian pointing accuracy of laser beams due to the low divergence nature of the high frequency optical wavelengths utilised. Laser communication terminals handle the strict pointing accuracy requirements by segmenting the pointing strategies into coarse and fine pointing mechanisms. The coarse pointing assembly (CPA), the focus of this research, provides large field of view, low bandwidth laser beam pointing with minimal static and dynamic pointing error. Common pointing budgets allocate errors up to 100-500 μrad for static pointing and 1-5 μrad for dynamic pointing biases during operation. The CPA operates on the simultaneous actuation of two axes, elevation and azimuth, to anticipate relative spacecraft motion for inter-satellite links and alternatively, to Earth based ground stations. For complete closed loop pointing of the CPA, the elevation axes power, motor control signals and sensor feedback needs to pass over the rotating azimuth axis. The state of the art mechanisms used to accomplish this are flex wraps, flexible PCBs coiled around a central axis, which unspools and retracts for azimuthal movements. However, flex wraps negatively impact CPA pointing performance through their material stiffness destabilisation torques, limit the CPA lifetime due to material fatigue characteristics, pose manufacturing and assembly complexity, inhibit fully continuous 360 degree pointing operation of the CPA, and restrict the instruments achievable slew rates. The development in this domain of CPA technology has remained largely stagnant for multiple years.

This research looks into the field of contactless technology to analyse, design and develop a novel wireless interface for the continuous rotation, low torque and high precision pointing of CPAs, by aiming to eliminate all wired connections between the rotation axes. The conceptual design, analysis implementation and manufacturing of a contactless interface concept was carried out through a 10 month period, bringing the technology from TRL 1 to TRL 4, concluding with the testing and demonstration of a fully synchronous wireless power and bi-directional data transmission interface for CPA applications. The system accomplished a DC-DC raw power transfer efficiency of 73% and a maximum full duplex symbol rate over the contactless link of 100 kbps. Therefore, the contactless CPA interface displays the ability for a miniaturised and fully integrated solution to wirelessly couple power over free space while simultaneously enabling data transfer over a single coil element independent of any rotation effects.

The test results of the manufactured contactless interface displayed promise for the future implementation in a CPA, the system enables the full, undisturbed, continuous rotation of the CPA during operation. Complete pointing performance testing was not performed, however it is hypothesised that the implementation of the system would eliminate line resistance and material stiffness hysteric effects, conceivably removing all flex wrap induced pointing destabilisations and providing zero wear operations, removing instrument lifetime restrictions. However In doing so, the power transfer efficiency and number of signal data lines suffer a reduction as compared to flex wraps. As a part of this research, space radiation and thermal effects were not assessed in detail, however their influence was found to impact the overall design direction and component choices.

The mini optical ground station (MISO), on top of the Aerospace Engineering faculty, serves the purpose of demonstrating Free Space Optical Communication technology. A downlink from low earth orbit imposes pointing requirements beyond what can be achieved by the telescope mount. A Fine Pointing Assembly (FPA), the topic of this thesis, aims to reduce residual pointing error to facilitate coupling into a single-mode fiber architecture. This thesis addresses the design, implementation, and characterization of FPA to fulfill this job in a lab environment. The FPA was built as a feedback control system actuating a fine steering mirror to counteract tip-tilt with feedback from a SWIR camera and the center of the gravity tracking algorithm. The bandwidth of this system is limited by latency from software implementations of the interfaces. The developed FPA removes 76% of the Tip-Tilt disturbance energy and can be improved by mitigation efforts toward latency reduction.

Future Very High-Throughput Satellites are foreseen to implement optical ground-to-satellite feeder links to achieve multi-terabit-per-second data rates. Optical links however, are highly susceptible to atmospheric losses caused by turbulence, absorption, and scattering, especially at low elevation angles. Despite prior confirmation of the feasibility of cloud-free network availability, high-latitude stations have been notably absent from optical feeder link studies due to the limitation of Geostationary orbits in providing sufficiently high link elevation angles. Tundra orbits present a promising alternative to Geostationary orbit, requiring two satellites to ensure uninterrupted coverage of high-latitude regions like Europe and Canada with link geometries highly suitable for optical communication. This paper addresses optical feeder link implementation aspects in Tundra orbits and selects a suitable orbit to service Canada while considering aspects such as coverage, radiation environment, pointing angles, and delta-v impact. End-to-end simulations, including downlink and uplink amplitude statistics, are presented to assess dynamic turbulence penalties. These simulations focus on angular anisoplanatism and highlight the link geometry impact on adaptive optics efficacy. This analysis anticipates a mean 4.1 and 4.7 dBm link budget advantage for Tundra configurations for two and three satellites respectively. These findings highlight Tundra constellations' potential to enhance satellite communication infrastructure, providing robust, efficient service in regions where Geostationary orbits faces limitations.

Since their first use during the Apollo 15, 16, and 17 missions, laser altimeters have become indispensable for planetary exploration, enabling topographic mapping of rocky bodies throughout the solar system. The latest European contributions in this field are represented by the BepiColombo Laser Altimeter (BELA) and the Ganymede Laser Altimeter (GALA), developed by the DLR Institute of Planetary Research and currently en route to Mercury and the Jovian moons, respectively. However, to date, such instruments have only been deployed on large satellites, failing to meet miniaturised systems' SWaP (Size, Weight, and Power) requirements.

This thesis investigates the adaptation of laser altimeter technology for smaller platforms, focusing on the NLA (New Laser Altimeter) developed for the SER3NE (Selene's Explorer for Roughness, Regolith, Resources, Neutrons, and Elements) mission proposal. The instrument aims to improve the precision of Lunar topographic data to support the characterisation of future landing sites for crewed missions from a 12 U microsatellite. To meet the stringent SWaP constraints, the design will feature a transceiver and a single-photon detection system — an approach never previously applied to topographic altimeters.

This thesis aims to develop an optical design that fits the instrument within a 3 U volume. Given the innovative nature of the design, a trade-off analysis was conducted to evaluate several configurations based on their compactness, cross-coupling between the laser source and detector, footprint size at the target, co-alignment between the transmitter and receiver, and transmission losses. The selected design stood out in providing a footprint size in the range allowed by the requirements, reducing the probability of damage to the detector due to internal reflections from the laser, and guaranteeing a more stable co-alignment between the transmitter and receiver paths. It includes a transmitter with two 45° folding mirrors, a borehole mirror that allows the laser to pass while deflecting the returning signal, and a shared telescope.

A test campaign is then conducted on a dedicated optomechanical design to assess 1) its compliance with some trade-off criteria and 2) its functionality. The expansion performance is demonstrated as predictable and reliable, guaranteeing the desired divergence and footprint at the target. On the other hand, the transmittance is increased above the constraints by varying the angle of incidence of the light on the band-pass filter. For the functional tests, the quality of the wavefront is first evaluated, resulting in an aberration-free transmitter and a slightly worse receiver, but still acceptable for altimetry applications. Finally, the prototype is aligned thoroughly and tested for timing measurements with single-pixel detectors, successfully providing the range to the target.

In the next instrument development phase, throughout the pre-phase A studies in collaboration with ESA (European Space Agency) and relevant companies, the prototype will be enhanced to verify the performance in the other criteria and test its functionality in a relevant environment with a flight campaign.

FSO satellite communication enables data transfer at high bandwidths, low latency, and high security. Due to these benefits, FSO communication services with satellite constellations can hugely improve global connectivity and bandwidth, while maintaining sufficient security. Modeling a global FSO satellite communication service is beneficial for gaining insight into, and performing preliminary analysis on such missions. However, these missions are complex problems with several physical processes occurring at different time scales. Macroscopic processes in the order of minutes to hours consist of relative platform dynamics (eg. aircraft and satellite motion) and atmospheric attenuation. Microscopic processes consist of atmospheric turbulence- and platform disturbances in the order of milliseconds and transporting photons/bits in the order of nanoseconds. To our knowledge, there are currently no end-to-end models that accurately and efficiently simulate a combination of all these processes. To overcome this, a multi-scale method is proposed that simulates and combines the physics of the microscopic and macroscopic processes of a global FSO satellite communication mission. A bi-directional air-to-space link is chosen as a use case. The proposed model simulates a 1.5 hours communication mission in 6 minutes with 8 Gb random access memory (RAM). Platform dynamics and jitter are the most dominant effects. Including microscopic processes results in a performance difference in signal power of 10 dB to 80 dB depending on the strength of the atmospheric channel. Using the multi-scale method, we can efficiently and realistically simulate an FSO communication service end-toend between an aircraft and a satellite constellation.

Laser satellite communications for ground-to-ground and ground-to space links is influenced by atmospheric turbulence. Typically, theoretical models are used to predict link performance and the influence of the atmosphere on variables such as beam wander, spot size and scintillation at the receiver. These variables are used as input for system design. However, the models, which predict the individual variables have not yet been empirically verified for their individual contributions. Therefore, this paper focuses on retrieving the individual variables estimated by theoretical models from measurements representing these variables. These measurements have been retrieved in a 10 km uplink path, ground-to-ground field test. In this test, the receiver plane is imaged. The spot location and size and the longitudinal, radial and on-axis scintillation indices are estimated from the measurements. Also, turbulence strength is measured along the path and are used as inputs for the theoretical models. For weak turbulence conditions the theory and test results fall within the same order of magnitude. For all effects, except on-axis scintillation, the values predicted by theory lie at a slightly higher value than retrieved with the field test for the weak turbulence regime. For strong turbulence conditions the difference in results between theory and data grows, up to order one. To the authors’ knowledge, this is the first time the turbulence variables have been experimentally investigated in so much detail. It is clear from the differences, that both the experimental setup and theory for horizontal ground-to-ground links need attention to close the gap between them, especially for turbulence conditions outside the Rytov approximation range.

The future will see an increase in missions headed for deep space. Advanced payloads will require higher data rates in more efficient spacecraft. Conventional space communication is facilitated through RF technology. Although it is very reliable, doubts arise on its future sustainability because of limited bandwidth, inefficiency and a saturated spectrum. Deep space optical communication (DSOC) can mitigate these issues and may present a sustainable future solution. In this thesis work, a concept is generated for the link and optomechanical design of a Deep space Optical Terminal (DOT), which is meant to be able of high data rate laser communication between the Moon and Earth. Emphasis lies on link analysis and design parameters required to realize the link. Subsystems and budgets are highlighted. The design process is discussed using systems engineering methods and tools.

The goal of this report is to outline the sub-system design of the local sensing system chosen as the final concept in [1], to satisfy the mission need statement: measure the atmospheric conditions with full three-dimensional coverage of a wind farm to optimize its operational performance and control. This statement is derived from the need to improve the control and performance of wind farms through more informed processes and decisions, a task that meteorological masts would usually take on. However, the providable coverage is very low in comparison to the one a UAV based system could provide. UAVs have the potential to significantly increase the measurement coverage around an entire wind farm and in turn return to the user more valuable data. To approach the finding of a solution to this problem, the project was divided into four: planning, concept definition, concept exploration and detailed design. From the first two phases came unique concepts exploring remote and local sensing options, combined with a range of UAV types including hybrid, fixed-wing and rotor. Through a detailed trade-off process and sensitivity analysis, the agreed upon final solution came to be a local sensing concept that makes use of many hybrid drones. In the fourth and final phase, where we now find ourselves, the detailed concept is unpacked and designed into a marketable system that is capable of satisfying the underlying MNS. In this stage the design was split into three design groups: UAV design, ground station design, swarm design.

Global internet traffic is subject to exponential growth. Free Space Optical communication can accommodate for this increase. Current communication systems prove to be capable of transferring Terabits/s of data per link to and from a satellite. This is done by coupling the laser beam into a fiber, which is connected to a computer capable of processing the data in the laser beam. The downside of these systems is that they require position sensors to accurately steer the beam into the fiber. These sensors limit the amount of light reaching the receiver and thereby limit data capacity. Furthermore, these sensors may drift due to thermal loads, causing the system to lose performance. By removing the position sensor and implementing the gradient-based algorithm Extremum Seeking Control, the same performance can be achieved without risk of fiber drift. Furthermore, this approach would reduce system complexity, relax tolerances imposed on these systems and most importantly, increase the average fiber coupling efficiency of satellite links by 40%. This is verified by a simulink model and breadboard experiment.

The resolution of ground-based telescopes is limited by the blurring caused by Earth’s turbulent atmosphere. Adaptive Optics (AO) is a technique used to correct atmospheric distortions, restoring images to diffraction-limited resolution. The main element of an AO system is the Deformable Mirror (DM), which compensates the distortions of the incoming wavefronts. There are several technologies available to actuate DMs. TNO has developed a novel DM equipped with electromagnetic actuators. Some of the design parameters of the DM have not yet been studied and are subject of optimization. A modeling approach is presented for studying the design sensitivities of the TNO DM. To increase optical performance, a study of optimal actuator layouts and mechanical parameters is shown. For the figure of merit in the optimization, a Fourier modal basis is used. Instead of the more commonly used Zernike basis from literature. This approach constitutes a blueprint for the design of electromagnetic DMs. Its use may be extended to other AO applications such as lithography or ophthalmology.