Keywords: In-space manufacturing and assembly
Abstract: Assembling large-scale infrastructure in space presents significant technical and logistical challenges. Traditionally, this process involves dividing complex systems into smaller modules, launching them separately, and assembling them in orbit—often relying on sophisticated and costly robotic systems.

This paper introduces a robotic push-based self-assembly approach to create arrays of hexagonal tiles. Such tiles and arrays are prevalent in space structures, and in particular in the upcoming space base solar power stations (SBSP) that reduces mechanical complexity by employing a passive "pusher" mechanism. The system enables autonomous alignment and propulsion of hexagonal tiles into desired configurations without the use of advanced robotics, offering a simple yet scalable geometric arrangement suitable for large-scale space structures.

Experimental validation was performed using a low-friction test-bed (an air hockey table) to approximate microgravity conditions. Results indicate that the pusher mechanism successfully enables directional self-assembly of hexagonal tiles in a linear configuration. Building on this outcome, future work will focus on combining multiple linear arrays to form larger, scalable hexagonal assemblies for space infrastructure applications.

Keywords: In-space manufacturing and assembly, From Earth to Space/Extreme Environments, Planetary exploration
Abstract: Large planar deployable lightweight structures, such as solar arrays and array antennas, with high packaging efficiency, will benefit future space missions. However, their two-dimensional compact stowage poses challenges due to the thin-film device's non-negligible thickness, which causes circumference differences when rolled up. This paper proposes a novel folding method, termed Yasuhara fold, which introduces localized foldbacks to mitigate circumference difference accumulation, enabling compact and uniform stowage even with thick, multilayer flexible substrates. Theoretical analysis and experimental results using prototypes demonstrate superior stowage efficiency and stable deployment characteristics compared to conventional approaches. The proposed folding method provides a promising foundation for scalable, multifunctional, autonomously deployable flexible planar structures aboard satellites as well as on the surfaces of celestial bodies.

Keywords: Manipulation in space, Orbital servicing and debris removal, In-space manufacturing and assembly
Abstract: Future space missions, including satellite life extension, orbital asset inspection, and deorbiting, rely heavily on space robotic manipulators. However, testing these robots on Earth presents significant challenges, as they are designed to operate in zero gravity but must be tested under 1 g conditions. Since the joints of most space manipulators cannot support their own weight under Earth gravity, mechanical support systems are required to reduce the gravitational joint loads. The Institute of Robotics and Mechatronics at the German Aerospace Center (DLR) and the University of Duisburg-Essen have developed the Motion Suspension System (MSS) - a cable-driven parallel robot that enables ground-based testing of space robotic manipulators in a fully three-dimensional workspace. This paper presents a quantitative assessment of force reconstruction accuracy, for ground-truth validation of the MSS. Using optical motion capture, we achieved sub-degree accuracy in measuring suspension force direction, identifying systematic errors and improving system performance by 15.6 %. This study is crucial to validate the MSS as a qualification platform for space robotic manipulators.

Keywords: Intelligent and autonomous space robotics systems
Abstract: Validating physics engines under microgravity conditions is critical for designing and controlling robotic systems in space, where friction, joint clearance, and contact can strongly influence system behavior. However, a lack of experimental data has limited efforts to validate and benchmark these simulators in flight-relevant environments. This work presents the design and execution of a free-floating microgravity experiment conducted on a parabolic flight to generate a dataset for validating rigid-body simulations. The experiment captures the centripetal deployment of three deployable mechanisms: a translational scissor, a polar scissor, and a pop-up extending truss (PET), using synchronized high-speed video and motion-capture sensors. We utilize multi-view reconstruction techniques to convert image data into 3D trajectory data of rigid elements across deployment. This dataset is used to validate the Dojo physics engine, which models full multi-rigid-body dynamics, including contact, joint friction, and clearance. By optimizing simulation parameters to match experimental trajectories, the accuracy of the model is quantified to assess sim-to-real alignment. Initial results show 1.3cm, 1.2cm, and 4.4cm RMSE for the scissor, polar scissor, and PET systems, respectively. This work contributes one of the first publicly documented microgravity datasets for evaluating rigid-body simulators and demonstrates a repeatable experimental framework for validating space robotics.

Keywords: Manipulation in space
Abstract: In the field of human spaceflight, humanoid robots that can interact with tools represent a valuable asset to long-term missions with transient human residence. To interact with their environment, due to form factor constraints, robots of this class feature tendon-driven robotic 'hands' as end effectors. One of the main failure points of robotic hands is the tendon terminations. The tendon terminations transfer the force of the grip and see high cyclic loading. Many studies focus on the max strength of these terminations but fail to study the performance over cyclic loading. This paper introduces a novel tendon termination that is compact, reliable, and repeatable. This is shown through repeated loading to a worst-case scenario lifting condition, as well as characterization of the ultimate strength of the tendon termination.

Keywords: Teleoperation and user interfaces, Multi-robot cooperation/collaboration, Intelligent and autonomous space robotics systems
Abstract: Remote assets generally have myopic sensor feedback that does not provide sufficient information alone to maintain situational awareness for effective teleoperation. Situational awareness has a large impact on mission outcomes: salient information fused and appropriately displayed to a remote operator has been shown to result in higher mission success. Additional robots or reconfigurable sensing platforms can be used to address this problem by providing additional views of the task. However, manually allocating and positioning additional resources to obtain requisite views will further burden remote operators. To address this issue, we have developed a framework to autonomously compute informative, supporting views of a task and dispatch resources to obtain the supporting views. The key insight is to leverage information in a task model called an Affordance Template that encodes important task objects and their relationship to the operated robot to predict informative viewing geometries. We demonstrate the system evaluating metrics to provide sensor coverage and to reduce target object pose uncertainty.

Keywords: Intelligent and autonomous space robotics systems, Multi-robot cooperation/collaboration, Planetary exploration
Abstract: Accurate localisation in planetary robotics enables the advanced autonomy required to support the increased scale and scope of future missions. The successes of the Ingenuity helicopter and multiple planetary orbiters lay the groundwork for future missions that use ground–aerial robotic teams. In this paper, we consider rovers using machine learning to localise themselves in a local aerial map using limited field-of-view monocular ground-view RGB images as input. A key consideration for machine learning methods is that real space data with ground-truth position labels suitable for training is scarce. In this work, we propose a novel method of localising rovers in an aerial map using cross-view-localising dual-encoder deep neural networks. We leverage semantic segmentation with vision foundation models and high volume synthetic data to bridge the domain gap to real images. We also contribute a new cross-view dataset of real-world rover trajectories with corresponding ground-truth localisation data captured in a planetary analogue facility, plus a high volume dataset of analogous synthetic image pairs. Using particle filters for state estimation with the cross-view networks allows accurate position estimation over simple and complex trajectories based on sequences of ground-view images.

Keywords: Planetary exploration, Intelligent and autonomous space robotics systems, From Earth to Space/Extreme Environments
Abstract: Robot navigation in unstructured environments requires multimodal perception systems that can support safe navigation. Multimodality enables the integration of complementary information collected by different sensors. However, this information must be processed by machine learning algorithms specifically designed to leverage heterogeneous data. Furthermore, it is necessary to identify which sensor modalities are most informative for navigation in the target environment. In Martian exploration, thermal imagery has proven valuable for assessing terrain safety due to differences in thermal behaviour between soil types. This work presents OmniUnet, a transformer-based neural network architecture for semantic segmentation using RGB, depth, and thermal (RGB-D-T) imagery. A custom multimodal sensor housing was developed using 3D printing and mounted on the Martian Rover Testbed for Autonomy (MaRTA) to collect a multimodal dataset in the Bardenas semi-desert in northern Spain. This location serves as a representative environment of the Martian surface, featuring terrain types such as sand, bedrock, and compact soil. A subset of this dataset was manually labeled to support supervised training of the network. The model was evaluated both quantitatively and qualitatively, achieving a pixel accuracy of 80.37% and demonstrating strong performance in segmenting complex unstructured terrain. Inference tests yielded an average prediction time of 673ms on a resource-constrained computer (Jetson Orin Nano), confirming its suitability for on-robot deployment. The software implementation of the network and the labeled dataset have been made publicly available to support future research in multimodal terrain perception for planetary robotics.

Keywords: Planetary exploration, Intelligent and autonomous space robotics systems, From Earth to Space/Extreme Environments
Abstract: Space missions shift their focus to the lunar surface, for which they require autonomous robots that work in extreme lighting conditions. Most existing datasets used for developing advanced Simultaneous Localisation And Mapping (SLAM) algorithms focus on urban environments like office buildings or cities. Only a limited number of datasets show deserted landscapes without vegetation, and they are usually captured during daylight. Simulations are frequently used in the absence of real or analogue data, but they have fidelity limitations compared to complex planetary environments. The LunaPolaris dataset addresses this issue by providing twelve sequences recorded in a lunar analogue facility using a small rover with a stereo camera, a solid-state LiDAR, and an IMU. These datasets can be used for validating the robustness of visual SLAM or Light Detecting And Ranging (LiDAR)-SLAM in challenging, lunar-like environments, highlighting the challenges around the extreme lighting conditions on the Moon. In this paper, we assess current gaps in existing datasets, present an overview of LunaPolaris, elaborate on the environment and technical details of how the data collection was conducted, and provide initial validation results using stereo and LiDAR SLAM.

Keywords: Planetary exploration, Intelligent and autonomous space robotics systems
Abstract: Precise localization of mobile robots is key to many (semi)-autonomous operations, such as planetary exploration. In situations, where Global Navigation Satellite Systems (GNSS) is unavailable, Ultra-Wideband (UWB) technology is a common replacement. This typically relies on the positions of distributed anchors to be known beforehand. In this work we expand on a system, that remotely distributes the anchors, which means the position of them is unknown. The robot is equipped with three UWB tags, which perform Two-Way-Ranging (TWR) distance measurements with all the anchors. These distances are used to determine the relative position of the anchors to the robot. These positions are interpreted as landmarks in an Extended Kalman Filter (EKF)-Simultaneous Localization and Mapping (SLAM) algorithm, which combines them with the wheel odometry of the robot. Our experiments show promising results in a setup with four anchors, performing considerably better than the wheel odometry on its own. The system is also capable of operating through an outage of the UWB anchors. After such an outage the pose of the robot is corrected in multiple experiments, though not to the same standard as before the outage.

Keywords: Intelligent and autonomous space robotics systems, Space robotic locomotion and terramechanics, Mission planning
Abstract: This paper introduces the Lunar Rover Mini (LRM), an open-source, low cost mobile robotic platform developed for space robotics research. Started in 2015 at DLR's Institute of Robotics and Mechatronics, the LRM is built using off-the-shelf hardware and 3D-printed components, with custom body and bogie PCBs that can be released and made openly available for reproduction. It features a modular software framework and robust perception capabilities that support autonomous functionality. The rover’s architecture includes a 6-wheel triple-rocker-bogie suspension system, a stereo RGB-D vision camera with an integrated IMU, and a 6-DOF robotic arm with a gripper. Middleware frameworks such as ROS, Simulink, and RAFCON are used to enable teleoperation, autonomous navigation, arm manipulation, and frontier-based exploration. Moreover, the compact rover measures 36x26x39 cm and has a total mass of 3.7 kg, while achieving a maximum speed of 0.13 m/s. The LRM has been successfully field-tested at different summer schools, demonstrating autonomous 3D mapping, frontier-based exploration, and robotic arm grasping while monitoring electric current.

Ongoing work includes the integration of a YOLOv7-based vision pipeline for object detection in mapped environment, integration of IMU data into the transformation tree for improved SLAM accuracy, and further mechanical enhancements. Also, increasing the complexity of autonomous exploration strategies through an utility function. The LRM seeks to bridge the accessibility gap in space robotics by offering a cost-effective, modular, easy-to-assemble rover that features capabilities comparable to those of state-of-the-art planetary rovers.

Keywords: Planetary exploration, Space robotic locomotion and terramechanics
Abstract: Permanently Shadowed Regions (PSRs) near the lunar poles are of interest for future exploration due to their potential to contain water ice and preserve geological records. Their complex, uneven terrain favors the use of legged robots, which can traverse challenging surfaces while collecting in-situ data, and have proven effective in Earth analogs, including dark caves, when equipped with onboard lighting. While exteroceptive sensors like cameras and lidars can capture terrain geometry and even semantic information, they cannot quantify its physical interaction with the robot—a capability provided by proprioceptive sensing. We propose a terrain mapping framework for quadruped robots which estimates elevation, foot slippage, energy cost, and stability margins from internal sensing during locomotion. These metrics are incrementally integrated into a multi-layer 2.5D gridmap that reflects terrain interaction from the robot’s perspective. The system is evaluated in a simulator that mimics a lunar environment, using the 21 kg quadruped robot Aliengo, showing consistent mapping performance under lunar gravity and terrain conditions.

Keywords: Mission planning, Intelligent and autonomous space robotics systems, Space logistics
Abstract: Accurate and robust Orbit Determination (OD) is foundational for numerous space robotics applications, including autonomous docking, planetary rover navigation, and aerial probe control. Traditional approaches such as Weighted Least Squares and the Extended Kalman Filter are widely used due to their statistical reliability, although they can suffer from sensitivity to poor choices of the initial conditions. Recent advances in Machine Learning, particularly Physics-Informed models, offer promising alternatives [1], [2], [3], [4], [5]. Building on earlier work applying Physics-Informed Extreme Learning Machines (PIELMs) to OD [6], [7], a similar framework is introduced here that incorporates orbital dynamics directly into the learning process. This approach is further extended with a Deep PIELM variant featuring autoencoder-based hidden layers [8] to enhance representational capacity and robustness. Comparative results demonstrate the benefits of these deeper architectures in improving OD performance while keeping training times low, positioning them as effective complements to classical techniques.

Keywords: Mission planning, Intelligent and autonomous space robotics systems, Space logistics
Abstract: In recent years, the demand for autonomous spacecraft operations has grown significantly, becoming a critical factor across all areas of space engineering. Accurate onboard orbit determination (OD) and orbit prediction (OP) are essential for reliable autonomous mission planning. However, onboard orbit modeling remains challenging due to inherent uncertainties and environmental complexity. This paper presents a hybrid approach combining traditional physics-based OD with a machine learning (ML) method inspired by Peng and Bai [1]. A lightweight Artificial Neural Network (ANN) is implemented for real-time onboard OP error correction. Simulations demonstrate the method’s effectiveness using realistic GPS tracking data generated by a high-fidelity model. Results show significant improvements in OP accuracy, reducing maximum along-track errors from approximately 5 km to about 1.2 km over a three-day OP horizon. The hybrid method is planned to be experimentally validated onboard the ATHENE-1 satellite [2], scheduled for launch in 2026, demonstrating enhanced spacecraft autonomy.

Keywords: Mission planning, Orbital servicing and debris removal, Prox ops/rendezvous/docking
Abstract: Autonomous mission planning for Active Debris Removal (ADR) must balance efficiency, adaptability, and strict feasibility constraints on fuel and mission duration. This work compares three planners for the constrained multi- debris rendezvous problem in Low Earth Orbit: a nominal Masked Proximal Policy Optimization (PPO) policy trained under fixed mission parameters, a domain-randomized Masked PPO policy trained across varying mission constraints for improved robustness, and a plain Monte Carlo Tree Search (MCTS) baseline. Evaluations are conducted in a high-fidelity orbital simulation with refueling, realistic transfer dynamics, and randomized debris fields across 300 test cases in nominal, reduced fuel, and reduced mission time scenarios. Results show that nominal PPO achieves top performance when conditions match training but degrades sharply under distributional shift, while domain-randomized PPO exhibits improved adaptabil- ity with only moderate loss in nominal performance. MCTS consistently handles constraint changes best due to online replanning but incurs orders-of-magnitude higher computation time. The findings underline a trade-off between the speed of learned policies and the adaptability of search-based methods, and suggest that combining training-time diversity with online planning could be a promising path for future resilient ADR mission planners.

Keywords: Orbital servicing and debris removal, Mission planning, Space logistics
Abstract: Active debris removal (ADR) is critical to prevent the continued growth of debris, with an annual removal rate of at least five debris to stabilize the debris population in Low Earth Orbit (LEO). This study addresses the removal of rocket bodies in Sun Synchronous Orbit regions using the orbital transfer vehicle (OTV) equipped with de-orbit kits, and supported by a space logistics mission including on-orbit refueling and resupply to prolong OTV’s lifetime. The ant colony optimization is employed to optimize the removal sequence for a removal mission of 25 and 35 debris. The sequences are evaluated across multiple OTV designs, the resulting Pareto-optimal solutions in mission time and propellant mass reveal the performance and efficiencies in propellant consumption of each design. The results provide valuable insights for mission designs to balance the fuel capacity and the servicing frequency, guiding more efficient and sustainable ADR mission planning.

Keywords: Intelligent and autonomous space robotics systems, Space logistics, From Earth to Space/Extreme Environments
Abstract: The German Aerospace Center (DLR), as part of the International Space Exploration Coordination Group (ISECG), shares the vision of sustainable human and robotic exploration of the Solar System. In this context, the EDEN LUNA project introduces a Moon-analogue greenhouse facility for the demonstration of nearly closed-loop bio-regenerative life support systems technology and plants cultivation for the purpose of feeding a crew. An autonomous robotic system EDEN Versatile End-effector (EVE) is to be integrated into the EDEN LUNA greenhouse, to partner with humans in support of this food production and to enable sustained extra-terrestrial exploration. EVE operates in a shared-autonomy manner, wherein an operator issues commands which trigger autonomous operation of robotic system. This is a highly significant feature which directly impacts the workload of astronauts inside the greenhouse. This preliminary study describes the design of EVE and compares EVE’s preliminary performance to existing studies on agricultural robotics. It also investigates space plant cultivation experiments and ground-based greenhouse analogues to compare them with the automatized scenario presented in this work.

Keywords: Manipulation in space, Intelligent and autonomous space robotics systems, Mission planning
Abstract: Robotic systems for intra-vehicular space operations must safely navigate and interact within confined, sensitive spacecraft environments. While traditional rigid free-flyers offer reliable mobility, they lack the adaptability and passive safety of soft robots. Soft continuum systems promise safer interactions and greater flexibility but remain underexplored in microgravity, where the absence of reaction forces makes their dynamics hard to model and control. To address this gap, we introduce FLOATS, a simulation framework for free-flying soft robots in microgravity. FLOATS facilitates realistic simulation of locomotion, manipulation, and planning for deformable on-orbit systems. We demonstrate grasping, free-floating object reconfiguration, and mobile manipulation tasks as a baseline for research on soft robotics planning and control in microgravity.

Keywords: Space robotic locomotion and terramechanics, Orbital servicing and debris removal
Abstract: The growing demand for space exploration has intensified interest in the performance and capabilities of space robotic systems. Traditional rigid space robotics are often hindered by limitations such as structural rigidity, limited adaptability, and suboptimal safety characteristics. In contrast, bioinspired robotic systems, constructed from compliant materials, offer significant advantages, including reduced mass, enhanced flexibility, and improved adaptability to unstructured environments. This paper presents a Nonlinear Model Predictive Control (NMPC) framework for trajectory tracking in a cable-driven soft space manipulator considering a free-floating scenario. The dynamic model of the soft robotic system was formulated using the Euler -Lagrange method to capture the coupled dynamics between the manipulator and the spacecraft base. Based on the derived model, an NMPC strategy was developed to achieve precise end effector trajectory tracking. Numerical simulations demonstrated the effectiveness of the proposed control approach, with a maximum deviation of less than 0.04 m and a final error on the order of 10^{-6}m. In addition, this paper presents a soft robot prototype and a preliminary grasping experiment. The grasping experiment, conducted on a flat surface under Earth’s gravity, demonstrates that it is capable of grasping a 1U CubeSat.

Keywords: Orbital servicing and debris removal, Manipulation in space, Prox ops/rendezvous/docking
Abstract: For non-destructive orbital debris capture, de-tumbling is often a necessary step. However, gaining control of objects that are large and rotating with an irregular or unpredictable trajectory is challenging. We present a concept and initial experiments aimed at establishing the feasibility of restraining such objects using a gecko-inspired adhesive attached to a tether: Gecko Lasso. We assume that the adhesive is brought into initial contact by an acquisition vehicle using a lightweight arm or extendable boom. As the object rotates, it wraps the tether around itself, rapidly increasing the maximum permissible tether tension. Experiments with rotating metal cylinders and a small sample of adhesive indicate that gentle attachment is possible for fast-moving targets; tangential velocities up to 1.3 m/s were tested with a 90% success rate. We conclude with a discussion of extensions to take this new concept to the next level of technology readiness.

Keywords: Orbital servicing and debris removal, Intelligent and autonomous space robotics systems, Manipulation in space
Abstract: This paper presents the design, implementation, and preliminary testing of IDRA, a modular inflatable robotic arm for In‑Space Servicing, Assembly, and Manufacturing (ISAM) applications. The system leverages inflatable link technology to achieve a high packing ratio and large operational workspace, enabling functionalities beyond the capabilities of conventional rigid manipulators. A planar prototype was developed for functional validation under laboratory conditions using visual servoing control. Experimental trials assessed positioning accuracy and repeatability across varying internal pressures, demonstrating reliable convergence to the target despite the inherent compliance of the inflatable structure. The results confirm the robustness of the control strategy. Building on these outcomes, future work will focus on integrating the latest chamber design with a deployment–retraction mechanism, validating it in thermal‑vacuum conditions, and advancing toward a fully operational multi‑DOF robotic system suitable for space environments.

Keywords: Orbital servicing and debris removal, Manipulation in space, Intelligent and autonomous space robotics systems
Abstract: The increasing population of uncooperative space debris in Low Earth Orbit (LEO) poses a significant risk to the sustainability of future space missions. The need for safe, robust and effective Active Debris Removal (ADR) strategies cannot be denied any more. This paper takes on the previous work on a small satellite novel hybrid-compliant two-stage mechanism, called the Soft Capture Unit (SCU), designed to enhance autonomous debris capture by enabling controlled energy-dissipative contact with flat-surfaced debris. This device must ensure enough contact time for its subsystem to be activated and adhere to the surface of the debris. To that extent, a dynamic state-space model of the designed system was developed, and an optimisation algorithm was implemented using experimental impact data collected in the University of Luxembourg’s Zero-G lab. By systematically varying the compliant system’s parameters, an optimised coefficient set of stiffness and damping was calculated to maximise contact duration under a representative impact velocity. Experimental tests emulating several mass factors show a contact time increase of at least +138.00 ms over a non-compliant configuration, significantly improving adhesion reliability and reducing risk-critical factors such as hard shocks for successful debris removal. These findings are validated through close-to-real scenarios testing, confirming the feasibility of the proposed mechanism for future ADR missions.

Keywords: Intelligent and autonomous space robotics systems, Prox ops/rendezvous/docking, Orbital servicing and debris removal
Abstract: This work presents a novel pipeline to recover the 3D structure of an unknown target spacecraft from a sequence of images captured during Rendezvous and Proximity Operations (RPO) in space. The target's geometry and appearance are represented as a 3D Gaussian Splatting (3DGS) model. However, learning 3DGS requires static scenes, an assumption in contrast to dynamic lighting conditions encountered in spaceborne imagery. The trained 3DGS model can also be used for camera pose estimation through photometric optimization. Therefore, in addition to recovering a geometrically accurate 3DGS model, the photometric accuracy of the rendered images is imperative to downstream pose estimation tasks during the RPO process. This work proposes to incorporate the prior knowledge of the Sun's position, estimated and maintained by the servicer spacecraft, into the training pipeline for improved photometric quality of 3DGS rasterization. Experimental studies demonstrate the effectiveness of the proposed solution, as 3DGS models trained on a sequence of images learn to adapt to rapidly changing illumination conditions in space and reflect global shadowing and self-occlusion.

Keywords: Orbital servicing and debris removal, Prox ops/rendezvous/docking, In-space manufacturing and assembly
Abstract: In this paper, we propose an online estimation method for the moment of inertia ratio (MIR) for uncooperative spacecraft using only externally obtained angular velocity measurements.In recent years, On-Orbit Servicing (OOS) has attracted significant attention as a next-generation satellite technology.To realize services involving physical contact, it is essential to estimate the target's inertial parameters in advance. The proposed method employs the Unscented Kalman Filter (UKF) to estimate the MIR solely from externally acquired angular velocity data. Through Monte Carlo simulations, the fundamental performance of the algorithm was validated, demonstrating that the MIR can be estimated with very high accuracy. Furthermore, we conducted simulations assuming that the observable angular velocity is limited to only two axes, and confirmed that the proposed method can still estimate the MIR with high accuracy even when the available information is restricted. This study provides a general framework applicable to uncooperative target with limited observation axes, contributing to the realization of next-generation OOS technologies.

Keywords: Orbital servicing and debris removal, Prox ops/rendezvous/docking, Intelligent and autonomous space robotics systems
Abstract: Pose estimation of uncooperative spacecraft, which is essential for on-orbit servicing, is challenging for highly axisymmetric targets such as rocket bodies. For these targets, accuracy is degraded because the rotation around the symmetric axis is difficult to observe. To address this challenge, this paper proposes a new CNN-based estimation framework that divides the problem into two stages. In the first stage, the framework estimates the rotation around the camera's line-of-sight to normalize the input image. In the second stage, it estimates the remaining pose components from this simplified image, thereby reducing the overall estimation complexity. Through evaluation experiments using custom datasets of both axisymmetric (H-IIA R/B) and nonaxisymmetric (SOHO satellite) spacecraft, we demonstrate that the proposed method significantly outperforms the accuracy of conventional single-stage estimation. Especially, for the axisymmetric target, the proposed method achieved a median absolute angular error of 0.59 degree, compared to the best-performing single-stage method result of 1.89 degree. These results indicate that an approach that progressively simplifies the estimation process is highly effective for the pose estimation of axisymmetric spacecraft, which includes the rotation about the symmetric axis.

Keywords: Planetary exploration, Intelligent and autonomous space robotics systems
Abstract: Building on our previous work with CNN-based architectures for Mars rover terrain classification, this paper investigates transformer-based semantic segmentation models. We benchmark three state-of-the-art transformers— SegFormer, Mask2Former, and UPerNet—against our best- performing CNN baseline, DeepLabV3+, using two Mars terrain datasets: AI4Mars and LabelMars. Evaluation metrics include pixel-level accuracy, mean Intersection over Union (mIoU), model parameter count, and inference speed (frames per second), among others. Our findings show that transformer models match or slightly outperform DeepLabV3+ on the smaller LabelMars dataset, despite limited training data. On the larger AI4Mars dataset, transformers significantly outperform DeepLabV3+, particularly in detecting rare but safety-critical terrain features such as the Big Rock obstacle class. However, DeepLabV3+ demonstrates better performance at lower image resolutions, highlighting the importance of resolution in fully leveraging the self-attention mechanisms of transformer-based models. To improve model robustness and mitigate overfitting, we incorporate semi-supervised learning using additional unlabeled AI4Mars images. In our efficiency analysis, SegFormer stands out by achieving the highest real-time inference throughput (FPS) while maintaining competitive mIoU, making it a strong candidate for onboard deployment—where both accuracy and computational efficiency are essential for autonomous navigation in future Martian missions. We further demonstrated the feasibility of multi-scale inference, showing that models trained at higher resolutions retain strong performance when deployed at lower resolutions, enabling efficient onboard computation. Overall, our results highlight the potential of transformer- based segmentation models to significantly enhance Mars rover autonomy by combining high segmentation accuracy with the real-time efficiency required for onboard systems.

Keywords: Intelligent and autonomous space robotics systems
Abstract: Deep space missions face extreme communication delays and environmental uncertainty that prevent real-time ground operations. To support autonomous science operations in communication-constrained environments, we present a partially observable Markov decision process (POMDP) framework that adaptively sequences spacecraft science instruments. We integrate a Bayesian network into the POMDP observation space to manage the high-dimensional and uncertain measurements typical of astrobiology missions. This network compactly encodes dependencies among measurements and improves the interpretability and computational tractability of science data. Instrument operation policies are computed offline, allowing resource-aware plans to be generated and thoroughly validated prior to launch. We use the Enceladus Orbilander’s proposed Life Detection Suite (LDS) as a case study, demonstrating how Bayesian network structure and reward shaping influence system performance. We compare our method against the mission’s baseline Concept of Operations (ConOps), evaluating both misclassification rates and performance in off-nominal sample accumulation scenarios. Our approach reduces sample identification errors by nearly 40%.

Keywords: Intelligent and autonomous space robotics systems, Planetary exploration, From Earth to Space/Extreme Environments
Abstract: The Martian Moons eXploration (MMX) is set to examine the Martian Moons Phobos and Deimos. As part of this mission, a small rover named IDEFIX will explore the surface of Phobos as the first in-situ system. The German Aerospace Center (DLR) develops the onboard autonomous navigation solution NAVDLR as a technology demonstration and also to increase the safety of rover driving operations. The harsh Phobos environment challenges the autonomous navigation due to its illumination conditions, the temperatures, and the radiation environment, especially as there remains a high uncertainty about the environmental parameters. This paper presents our approach to simulate varying and challenging environment effects on existing, representative data sets. We analyze the robustness of our autonomous navigation solution in relation to these conditions. Our results demonstrate that a loss of tracking can occur for challenging scenarios, but NAVDLR generally provides accurate and robust results. Finally, we propose mitigation techniques to reduce loss of tracking risk and improve navigation robustness.

Keywords: Intelligent and autonomous space robotics systems, Planetary exploration, From Earth to Space/Extreme Environments
Abstract: We present a fully decentralized routing framework for multi-robot exploration missions operating under the constraints of a Lunar Delay-Tolerant Network (LDTN). In this setting, autonomous rovers must relay collected data to a lander under intermittent connectivity and unknown mobility patterns. We formulate the problem as a Partially Observable Markov Decision Problem (POMDP) and propose a Graph Attention-based Multi-Agent Reinforcement Learning (GAT-MARL) policy that performs Centralized Training, Decentralized Execution (CTDE). Our method relies only on local observations and does not require global topology updates or packet replication, unlike classical approaches such as shortest path and controlled flooding-based algorithms. Through Monte Carlo simulations in randomized exploration environments, GAT-MARL provides higher delivery rates, no duplications, and fewer packet losses, and is able to leverage short-term mobility forecasts-offering a scalable solution for future space robotic systems for planetary exploration, as demonstrated by successful generalization to larger rover teams.

Keywords: Intelligent and autonomous space robotics systems, Multi-robot cooperation/collaboration, Manipulation in space
Abstract: Modular reconfigurable robots, consisting of interchangeable modules, enable structural adaptation to meet task-specific requirements in space exploration. However, the combinatorial growth of possible morphologies renders training a unified control policy significantly complicated. This work presents a decentralized reinforcement learning (Dec-RL) approach in which each fundamental robotic module is trained with an independent policy. The wheel modules employ Soft Actor-Critic (SAC) for locomotion, while the 7-DoF limbs use Proximal Policy Optimization (PPO) for steering and manipulation tasks, enabling zero-shot policy generalization in previously unseen configurations. During training and evaluation with digital twin simulation, the steering policy achieved high steering angle accuracy between the desired and induced steering angles with a Mean Absolute Error of 3.63 degrees. The manipulation policy reached 84.6% convergence with the intended offset to the target. The wheel policy reduced the average motor torque by 95.4% while achieving a 99.6% success rate. Hardware experiments in lunar analogue field validated zero-shot integration of these policies for autonomous locomotion, steering, and preliminary alignment for reconfiguration of our modular robots. The system switched smoothly between synchronous, parallel, and sequential modes without idle states or control conflicts, showing that the approach is scalable, reusable, and robust for modular lunar robots.

Keywords: Intelligent and autonomous space robotics systems, Planetary exploration, Space robotic locomotion and terramechanics
Abstract: Reliable autonomous navigation across the unstructured terrains of distant planetary surfaces is a critical enabler for future space exploration. However, the deployment of learning-based controllers is hindered by the inherent sim-to-real gap, particularly for the complex dynamics of wheel interactions with granular media. This work presents a complete sim-to-real framework for developing and validating robust control policies for dynamic waypoint tracking on such challenging surfaces. We leverage massively parallel simulation to train reinforcement learning agents across a vast distribution of procedurally generated environments with randomized physics. These policies are then transferred zero-shot to a physical wheeled rover operating in a lunar-analogue facility. Our experiments systematically compare multiple reinforcement learning algorithms and action smoothing filters to identify the most effective combinations for real-world deployment. Crucially, we provide strong empirical evidence that agents trained with procedural diversity achieve superior zero-shot performance compared to those trained on static scenarios. We also analyze the trade-offs of fine-tuning with high-fidelity particle physics, which offers minor gains in low-speed precision at a significant computational cost. Together, these contributions establish a validated workflow for creating reliable learning-based navigation systems, marking a substantial step towards deploying autonomous robots in the final frontier.

Keywords: Satellite formation-flying, Multi-robot cooperation/collaboration, Satellite swarms
Abstract: Integrating high-fidelity spacecraft simulators with modular robotics frameworks remains a challenge for autonomy development. This paper presents a lightweight, open-source communication bridge between the Basilisk astrodynamics simulator and the Robot Operating System 2 (ROS 2), enabling real-time, bidirectional data exchange for spacecraft control. The bridge requires no changes to Basilisk's core and integrates seamlessly with ROS 2 nodes. We demonstrate its use in a leader–follower formation flying scenario using nonlinear model predictive control, deployed identically in both simulation and on the ATMOS planar microgravity testbed. This setup supports rapid development, hardware-in-the-loop testing, and seamless transition from simulation to hardware. The bridge offers a flexible and scalable platform for modular spacecraft autonomy and reproducible research workflows.

Keywords: Satellite swarms, Satellite formation-flying, Prox ops/rendezvous/docking
Abstract: This paper proposes a novel six-degree-of-freedom (6-DOF) proximity control method using electromagnetic forces, which is required for realizing future on-orbit infrastructure. The long-duration stable and accurate proximity control method is required for on-orbit infrastructure such as on-orbit servicing and distributed aperture systems. Fuel-less electromagnetic forces enable such proximity control. Conventional electromagnetic proximity control faces two main challenges. The first is the trade-off between control accuracy and computation time. Rigorously calculating electromagnetic interactions improves accuracy but requires substantial computation time, while simplified models reduce computation time at the cost of accuracy. The second challenge is control singularities, where the electromagnetic accutuation matrix becomes rank-deficient. These conditions occur unexpectedly in the proximity region and compromise system safety. The proposed control architecture addresses both challenges. A computationally efficient model that accounts for coil arrangement effects—significant in close proximity—is combined with a State-Dependent Riccati Equation (SDRE) controller to be robust against singularities. A disturbance observer is also introduced to compensate for near-field modeling errors and enhance accuracy. Numerical simulations of docking maneuvers demonstrate the effectiveness of the proposed method. Even in scenarios where conventional methods fail due to singularities, the proposed method　achieved 1 cm position and 3° attitude accuracy while being nearly 10 times more computationally efficient. Case studies further confirm robustness across various realistic coil arrangements. Overall, the results indicate a practical framework for electromagnetic proximity control, combining computational efficiency with accuracy and robustness.

Keywords: Satellite swarms, Intelligent and autonomous space robotics systems, Fractionated spacecraft and constellations
Abstract: Multi-agent architectures for space missions have become increasingly popular due to the operational benefits provided by the ability to fractionate functionality between agents and gracefully scale the number of agents up or down to meet mission needs. The control of multi-agent satellite swarms through the interaction of magnetic fields generated by electromagnetic coils, known as electromagnetic formation flight (EMFF), has been previously explored and offers operational benefits such as mitigated reliance on finite propellant and non-mechanical coupling. However, the highly-coupled, stiff, and chaotic dynamics of EMFF can make control challenging. Current state-of-the-art methods rely on solving the dipole inversion subproblem using a computationally expensive nonlinear program at each control iteration. In this work, we present an approximate reformulation of the dynamics which addresses some of their problematic aspects. Using our dynamics, we formulate EMFF swarm control as a Markov decision process and use deep reinforcement learning to obtain an end-to-end policy for control. We apply this policy to a heterogeneous swarm architecture for EMFF based around a prime spacecraft with auxiliary inertial propulsion modalities. To our knowledge, this work demonstrates the first learned solution to dipole inversion, providing a constant time evaluation for the control loop. Our trained policy achieves sub-meter accuracy in position control for a small formation of two worker spacecraft.

Keywords: Intelligent and autonomous space robotics systems, Manipulation in space
Abstract: In this study, we apply Soft Actor Critic (SAC), a deep reinforcement learning algorithm, to achieve large-angle propellant-free attitude control of a solar sail under solar radiation pressure. Our solar sail model consists of a main square panel, a control boom attached to the center of the square panel via a two-axis gimbal mechanism, and two control vanes attached to the diagonal vertices of the square panel. We train motions of the boom and vanes with respect to the square panel via SAC so that the solar sail can change its attitude autonomously by a large angle (e.g. π rad) about an arbitrary axis with less work, less time, and higher precision. The results indicate that the trained model primarily exploits the coupling between internal motions due to the control boom and attitude motions, which we call the geometric attitude change, to achieve the goal. The trained model also exploits solar radiation pressure torques and the resulted total angular momenta to change the attitude efficiently, which we call the dynamic attitude change. The control boom and vanes are also effective in unloading total angular momentum exploiting solar radiation pressure torques.

Keywords: Intelligent and autonomous space robotics systems
Abstract: This paper presents a comprehensive analysis of how excitation design influences the identification of the inertia properties of rigid nano- and micro-satellites. We simulate nonlinear attitude dynamics with reaction-wheel coupling, actuator limits, and external disturbances, and excite the system using eight torque profiles of varying spectral richness. Two estimators are compared, a batch Least Squares method and an Extended Kalman Filter, across three satellite configurations and time-varying inertia scenarios. Results show that excitation frequency content and estimator assumptions jointly determine estimation accuracy and robustness, offering practical guidance for in-orbit adaptive inertia identification by outlining the conditions under which each method performs best. The code is provided as open-source.

Keywords: Satellite swarms, Fractionated spacecraft and constellations, Satellite formation-flying
Abstract: This study experimentally validates the principle of large-scale satellite swarm control through learning-aided magnetic field interactions generated by satellite-mounted magnetorquers. This actuation presents a promising solution for the long-term formation maintenance of multiple satellites and has primarily been demonstrated in ground-based testbeds for two-satellite position control. However, as the number of satellites increases beyond three, fundamental challenges coupled with the high nonlinearity arise: 1) nonholonomic constraints, 2) underactuation, 3) scalability, and 4) computational cost. Previous studies have shown that time-integrated current control theoretically solves these problems, where the average actuator outputs align with the desired command, and a learning-based technique further enhances their performance. Through multiple experiments, we validate critical aspects of learning-aided time-integrated current control: (1) enhanced controllability of the averaged system dynamics, with a theoretically guaranteed error bound, and (2) decentralized current management. We design two-axis coils and a ground-based experimental setup utilizing an air-bearing platform, enabling a mathematical replication of orbital dynamics. Based on the effectiveness of the learned interaction model, we introduce NODA-MMH (Neural power-Optimal Dipole Allocation for certified learned Model-based Magnetically swarm control Harness) for model-based power-optimal swarm control. This study complements our tutorial paper on magnetically actuated swarms for the long-term formation maintenance problem.

Keywords: Fractionated spacecraft and constellations, Satellite formation-flying, Mission planning
Abstract: Simulating the full set of satellite subsystems with high fidelity is essential for both mission design and operational planning. However, conventional simulators are often computationally intensive and difficult to scale, particularly in the context of distributed space systems (DSS) such as constellations, swarms, or formation-flying architectures. This work explores the use of machine learning (ML) as a lightweight, data-driven alternative for modeling the power subsystem, specifically, the prediction of solar array power output, battery input power, and battery output power. Using open-access telemetry from the BEESAT-4 CubeSat mission, Multilayer Perceptron (MLP) models were trained to estimate these quantities based on a wide range of onboard sensor data. The models achieved high accuracy, with mean absolute errors below 2% of the respective power ranges. Permutation Feature Importance analysis revealed that subsystem activity indicators, such as charger currents, sun vector orientation, and communication system states, play a critical role in power behavior. These findings demonstrate the feasibility of using ML to approximate subsystem dynamics with minimal computational overhead, and provide insights into sensor prioritization for future Digital Twin (DT) implementations in space systems.