With the rapid development of generative models and multimodal content editing technologies, the key challenge faced by synthetic image detection (SID) lies in cross-distribution generalization to unknown generation sources. In recent years, visual foundation models (VFM), which acquire rich visual priors through large scale image-text alignment pretraining, have become a promising technical route for improving the generalization ability of SID. However, existing VFM-based methods remain relatively coarse-grained in their adaptation strategies. They typically either directly use the final layer representations of VFM or simply fuse multi layer features, lacking explicit modeling of the optimal representational hierarchy for transferable forgery cues. Meanwhile, although directly fine-tuning VFM can enhance task adaptation, it may also damage the cross-modal pretrained structure that supports open-set generalization. To address this task specific tension, we reformulate VFM adaptation for SID as a joint optimization problem: it is necessary both to identify the critical representational layer that is more suitable for carrying forgery discriminative information and to constrain the disturbance caused by task knowledge injection to the pretrained structure. Based on this, we propose I2P, an SID framework centered on intrinsic importance perception. I2P first adaptively identifies the critical layer representations that are most discriminative for SID, and then constrains task-driven parameter updates within a low sensitivity parameter subspace, thereby improving task specificity while preserving the transferable structure of pretrained representations as much as possible.

Spatiotemporal vector retrieval has emerged as a critical paradigm in modern information retrieval, enabling efficient access to massive, heterogeneous data that evolve over both time and space. However, existing spatiotemporal retrieval methods are often extensions of conventional vector search systems that rely on external filters or specialized indices to incorporate temporal and spatial constraints, leading to inefficiency, architectural complexity, and limited flexibility in handling heterogeneous modalities. To overcome these challenges, we present a unified spatiotemporal vector retrieval framework that integrates temporal, spatial, and semantic cues within a coherent similarity space while maintaining scalability and adaptability to continuous data streams. Specifically, we propose (1) a Rotary-based Unified Encoding Method that embeds time and location into rotational position vectors for consistent spatiotemporal representation; (2) a Circular Incremental Update Mechanism that supports efficient sliding-window updates without global re-encoding or index reconstruction; and (3) a Weighted Interest-based Retrieval Algorithm that adaptively balances modality weights for context-aware and personalized retrieval. Extensive experiments across multiple real-world datasets demonstrate that our framework substantially outperforms state-of-the-art baselines in both retrieval accuracy and efficiency, while maintaining robustness under dynamic data evolution. These results highlight the effectiveness and practicality of the proposed approach for scalable spatiotemporal information retrieval in intelligent systems.

Artificial Neural Networks (ANNs) are powerful tools for complex decision-making tasks. While existing activation mechanisms often promote sparsity through thresholding, they lack explicit awareness of feature channel relevance, causing networks to continuously suffer from interference by noisy channels. Such irrelevant activation signals can propagate through the network and adversely affect the final decision. Inspired by observations that channel relevance can be reflected in both intrinsic activity levels and extrinsic decision weights, and that there is strong consensus between these two aspects, we propose AIEC (Activation with Intrinsic-Extrinsic Consensus), a novel activation mechanism that has the ability to identify and suppress irrelevant feature channels during training. With a basic threshold activation, AIEC leverages an intrinsic Activation-Counting Unit that tracks channel activation statistics, an extrinsic Decision-Making Unit that learns channel decision weights, and a Consensus Gatekeeping Unit that suppresses irrelevant channels based on the agreement between intrinsic and extrinsic channel relevance assessments. Extensive experiments demonstrate that AIEC can effectively suppress irrelevant channels and encourage sparser representations. Furthermore, AIEC is compatible with a wide range of mainstream ANN architectures and achieves superior performance compared to existing activation mechanisms across multiple tasks and domains.

Fraudulent activities have caused substantial negative social impacts and are exhibiting emerging characteristics such as intelligence and industrialization, posing challenges of high-order interactions, intricate dependencies, and the sparse yet concealed nature of fraudulent entities. Existing graph fraud detectors are limited by their narrow "receptive fields", as they focus only on the relations between an entity and its neighbors while neglecting longer-range structural associations hidden between entities. To address this issue, we propose a novel fraud detector based on Graph Path Aggregation (GPA). It operates through variable-length path sampling, semantic-associated path encoding, path interaction and aggregation, and aggregation-enhanced fraud detection. To further facilitate interpretable association analysis, we synthesize G-Internet, the first benchmark dataset in the field of internet fraud detection. Extensive experiments across datasets in multiple fraud scenarios demonstrate that the proposed GPA outperforms mainstream fraud detectors by up to +15% in Average Precision (AP). Additionally, GPA exhibits enhanced robustness to noisy labels and provides excellent interpretability by uncovering implicit fraudulent patterns across broader contexts. Code is available at https://github.com/horrible-dong/GPA.

Introduction: Lunar exploration is driven by a number of science and exploration goals (e.g., LEAG, 2016, 2017) [1]. One is determining the presence of water ice deposits in permanently shadowed regions (PSRs) [e.g., 2-5]. They are of scientific interest because past lunar exospheric conditions may be preserved in the ice [6], as well as for in-situ resource utilization.Multiple lines of evidence indicate that water ice is or may be present within some PSRs [e.g., 2, 6,7]. However, its areal distribution is largely unknown, particularly at sub-km spatial scales. Detecting surficial water ice within these PSRs could be achieved through active reflectance spectroscopy [e.g., 8].The VMMO Remote Sensing Payload: The proposed VMMO mission, which recently completed a CSA-funded Phase 0 and ESA-funded Phase A study, is intended to probe PSRs at spatial scales of metres to tens of metres. Active sensing will be accomplished via a three-band lidar system using wavelengths of 532, 1064, and 1560 nm [9].The selection of these wavelengths was designed to enable discrimination of water ice from mare and highlands. Highland regolith spectra are generally moderately bright in the visible region, flat to red-sloped beyond the visible region, usually with a weak plagioclase feldspar absorption band in the 1300 nm region, and sometimes with weak mafic silicate absorption bands in the 1000 and 2000 nm regions [10] (Figure 1). Mare regolith spectra are darker in the visible region, red-sloped beyond the visible region, and with weak to moderate mafic silicate absorption bands in the 1000 and 2000 nm regions (Figure 2). Water ice spectra are bright in the visible region, with blue sloped spectra beyond this region, and increasingly strong water ice absorption bands in the 1000, 1500, and 2000 nm regions [11] (Figure 3). At the VMMO wavelengths, these types of materials can be discriminated using both absolute reflectance and reflectance ratios for these three wavelengths.However, dust cover, percentage of ice covered by regolith, and regolith: ice ratio, and how dust and ice are mixed together could all influence the efficiency of detecting water ice. We have conducted laboratory experiments to test for how physical properties of ice + powdered lunar rock affect our ability to detect water ice using the three-band lidar system. We considered the following parameters: (1) different water ice: lunar material ratios in both intimate and areal mixtures; (2) local slope; and (3) different thicknesses of dust cover over water ice.Methods: Reflectance spectra (350-2500 nm) were acquired with an ASD Fieldspec Pro HR spectrometer. To simulate a lidar, we used a bifurcated fiber optic bundle, which provided co-aligned incidence and emission (i=e=0°). To measure the effects of local slope on lidar return, the samples were positioned at 10˚, 20˚, 30˚ and 40˚ off normal. All spectra were measured relative to a calibrated Spectralon panel.Results: Ice detection is possible using reflectance spectroscopy at 532, 1064, and 1530 nm for water ice abundances as low as 1 wt.%. We can determine or constrain whether water ice is exposed at the lunar surface, or covered by a thin dust layer. Both absolute and reflectance ratios using all three bands are required to fully detect and discriminate mare, highland, and water ice and to derive water ice surficial abundance.Water ice detection and discrimination is reliant on reflectance of the 1560 nm band, as this is where a strong water ice O-H overtone occurs, and reflectance in this region rapidly decreases with increasing ice abundance. In all cases, lunar regolith spectra are red-sloped (reflectance increasing toward longer wavelengths), and absolute reflectance varies with factors such as maturity and ilmenite abundance. Detection of water ice will be enhanced by comparing spectra acquired during a scan across a PSR, where mineralogical variations inside and immediately outside a PSR should be similar but vary in temperature [12, 13].Ilmenite detection: VMMO can also operate in passive reflectance mode. A portion of the detector will be equipped with a bandpass filter to measure reflected light in the ultraviolet (~290 nm) region. Ilmenite discrimination is best accomplished using an ultraviolet: visible reflectance ratio [14] (Fig. 4).Summary: VMMO provides an opportunity to search for surficial water ice at high sensitivity and spatial resolution useful for targeting locations for investigation by surface landers with precision guidance capabilities.Acknowledgements: This study has been supported by ESA, CSA, CFI, MRIF, NSERC, and UWinnipeg.References: [1] LEAG, (2016, 2017). https://www.lpi.usra.edu/leag/. [2] Nozette S. et al. (2001) JGR, 106, 23253–23266. [3] Lawrence, D.J. (2011). Nature Geosci., 4, 586-588. [4] Lawrence, D.J. (2017) JGR, 122, 21-52. [5] Lucey, P.G. (2009) Elements, 5, 41-46. [6] Feldman W. C. et al. (2001) JGR, 106, 23231–23251. [7] Colaprete, A., et al. (2010) Science, 330, 463-468. [8] Yoldi Z. et al. (2018) LPSC 49, # 2083. [9] Kruzelecky R. V. et al. (2018) ICES, 227, 1–20. [10] Pieters, C.M. (1986) Rev. Geophys.. 44, 557-578. [11] Clark, R.N. (1981) JGR, 86, 3087-3096. [12] Watson, K., et al. (1961) JGR, 66, 1598–1600. [13] Vasavada, A. R., et al. (1999) Icarus, 141,179–193. [14] Robinson, M.S., et al. (2007) GRL, 34, L13203. [14] C. Pitcher, et al. (2016) ASR, 57(5), 1197–1208. Figure 1. Reflectance spectra of two Apollo 16 highland regolith samples (

With the rapid development of deep learning, the increasing complexity and scale of parameters make training a new model increasingly resource-intensive. In this paper, we start from the classic convolutional neural network (CNN) and explore a paradigm that does not require training to obtain new models. Similar to the birth of CNN inspired by receptive fields in the biological visual system, we draw inspiration from the information subsystem pathways in the biological visual system and propose Model Disassembling and Assembling (MDA). During model disassembling, we introduce the concept of relative contribution and propose a component locating technique to extract task-aware components from trained CNN classifiers. For model assembling, we present the alignment padding strategy and parameter scaling strategy to construct a new model tailored for a specific task, utilizing the disassembled task-aware components. The entire process is akin to playing with LEGO bricks, enabling arbitrary assembly of new models, and providing a novel perspective for model creation and reuse. Extensive experiments showcase that task-aware components disassembled from CNN classifiers or new models assembled using these components closely match or even surpass the performance of the baseline, demonstrating its promising results for model reuse. Furthermore, MDA exhibits diverse potential applications, with comprehensive experiments exploring model decision route analysis, model compression, knowledge distillation, and more.

Due to the inability to receive signals from the Global Navigation Satellite System (GNSS) in extreme conditions, achieving accurate and robust navigation for Unmanned Aerial Vehicles (UAVs) is a challenging task. Recently emerged, vision-based navigation has been a promising and feasible alternative to GNSS-based navigation. However, existing vision-based techniques are inadequate in addressing flight deviation caused by environmental disturbances and inaccurate position predictions in practical settings. In this paper, we present a novel angle robustness navigation paradigm to deal with flight deviation in point-to-point navigation tasks. Additionally, we propose a model that includes the Adaptive Feature Enhance Module, Cross-knowledge Attention-guided Module and Robust Task-oriented Head Module to accurately predict direction angles for high-precision navigation. To evaluate the vision-based navigation methods, we collect a new dataset termed as UAV_AR368. Furthermore, we design the Simulation Flight Testing Instrument (SFTI) using Google Earth to simulate different flight environments, thereby reducing the expenses associated with real flight testing. Experiment results demonstrate that the proposed model outperforms the state-of-the-art by achieving improvements of 26.0% and 45.6% in the success rate of arrival under ideal and disturbed circumstances, respectively.

Due to its powerful representational capabilities, Transformers have gradually become the mainstream model in the field of machine vision. However, the vast and complex parameters of Transformers impede researchers from gaining a deep understanding of their internal mechanisms, especially error mechanisms. Existing methods for interpreting Transformers mainly focus on understanding them from the perspectives of the importance of input tokens or internal modules, as well as the formation and meaning of features. In contrast, inspired by research on information integration mechanisms and conjunctive errors in the biological visual system, this paper conducts an in-depth exploration of the internal error mechanisms of Transformers. We first propose an information integration hypothesis for Transformers in the machine vision domain and provide substantial experimental evidence to support this hypothesis. This includes the dynamic integration of information among tokens and the static integration of information within tokens in Transformers, as well as the presence of conjunctive errors therein. Addressing these errors, we further propose heuristic dynamic integration constraint methods and rule-based static integration constraint methods to rectify errors and ultimately improve model performance. The entire methodology framework is termed as Transformer Doctor, designed for diagnosing and treating internal errors within transformers. Through a plethora of quantitative and qualitative experiments, it has been demonstrated that Transformer Doctor can effectively address internal errors in transformers, thereby enhancing model performance.

Fraud detection has increasingly become a prominent research field due to the dramatically increased incidents of fraud. The complex connections involving thousands, or even millions of nodes, present challenges for fraud detection tasks. Many researchers have developed various graph-based methods to detect fraud from these intricate graphs. However, those methods neglect two distinct characteristics of the fraud graph: the non-additivity of certain attributes and the distinguishability of grouped messages from neighbor nodes. This paper introduces the Dynamic Grouping Aggregation Graph Neural Network (DGA-GNN) for fraud detection, which addresses these two characteristics by dynamically grouping attribute value ranges and neighbor nodes. In DGA-GNN, we initially propose the decision tree binning encoding to transform non-additive node attributes into bin vectors. This approach aligns well with the GNN’s aggregation operation and avoids nonsensical feature generation. Furthermore, we devise a feedback dynamic grouping strategy to classify graph nodes into two distinct groups and then employ a hierarchical aggregation. This method extracts more discriminative features for fraud detection tasks. Extensive experiments on five datasets suggest that our proposed method achieves a 3% ~ 16% improvement over existing SOTA methods. Code is available at https://github.com/AtwoodDuan/DGA-GNN.

Graph classification is a classic problem with practical applications in many real-life scenes. Existing graph neural networks, including GCN, GAT, and GIN, are proposed to extract useful features from complex graph structures. However, most existing methods’ feature extraction and aggregation inevitably mix the useful and redundant features, which will disturb the final classification performance. In this paper, to handle the above drawback, we put forward the Local Structural Separation Hypergraph Convolutional Neural Network (LoSS) based on two discoveries: most graph classification tasks only focus on a few groups of adjacent nodes, and different categories have their specific high response bits in graph embeddings. In LoSS, we first decouple the original graph into different hypergraphs and aggregate the features in each substructure, which aims to find useful features for the final classification. Next, the low-correlation feature suppression strategy is devised to suppress the irrelevant node-level and bit-level features in the forward inference process, effectively reducing the disturbance of redundant features. Experiments on five datasets show that the proposed LoSS can effectively locate and aggregate useful hypergraph features and achieve SOTA performance compared with existing methods.

Crawler detection is always an important research topic in network security. With the development of web technology, crawlers are constantly updating and changing, and their types are becoming diverse. The diversity and dynamics of crawlers pose significant challenges for feature applicability and model robustness. Existing crawler detection methods can only detect a limited number of crawlers by predefined rules and can not cover all types of crawlers; worse, they can be completely invalidated by the emergence of new types of crawlers. In this paper, we propose a reinforcement learning based web crawler detection method for diversity and dynamics (WC3D), which is composed of a feature selector and a session classifier. The feature selector selects the appropriate feature set for different types of crawlers with deep deterministic policy gradient. The session classifier makes crawler detection and provides rewards to the feature selector. The two modules are trained jointly to optimize the feature selection and session classification processes. Extensive experiments demonstrate the existence of crawler diversity and that the proposed method is still highly robust against the new type of crawlers and achieves state-of-the-art performance even without considering the dynamics of the crawlers.

A Robotic Active Debris Removal (ADR) mission requires close-proximity operations for capturing potentially tumbling, uncooperative targets and disposing of them. These operations rely on accurate relative navigation between the chaser and the target to safely perform the mission. In the case of a known piece of debris that has not been designed to be serviced or disposed of, the knowledge of its shape factor allows relative pose determination. In the case of unknown space debris, however, no relative pose determination is possible as no reference is available beforehand. In this paper, we derive an onboard software architecture for estimating the shape and dynamical model of an unknown, uncooperative space debris, and perform tracking, using features detected in the image and the associated depth. Firstly, an Error-State Kalman filter (ESKF) is used to estimate the chaser trajectory and attitude as well as the target shape in an arbitrarily fixed target frame. The next step in the architecture is to apply a Gaussian Processes (GPs) algorithm to model the rotational dynamics of the debris, allowing an accurate attitude propagation independently of its tumbling mode. Finally, a second ESKF is instantiated using existing knowledge to track the position and attitude of the target. This work assumes the pre-existence of a feature detection, tracking and matching algorithm, and fusing data from a depth sensor allowing the retrieval of pixel coordinates and depth for each detected feature. The proposed architecture has been tested and validated in a computer-simulated environment with parameters representative of a real-world ADR mission. Results from over 100 simulations show a mean position and attitude error of 0.3 m and 1.5 degrees, respectively.

The Enhanced Model Reference Adaptive Control (EMRAC) algorithm, augmenting the MRAC strategy with adaptive integral and adaptive switching control actions, is an effective solution to impose reference dynamics to plants affected by parameter uncertainties, unmodeled dynamics and disturbances. However, the design of the EMRAC solutions has so far been limited to single-input systems. To cover the gap, this paper presents two extensions of EMRAC to multi-input systems. The adaptive mechanism of both solutions includes the sigma$$ \sigma $$-modification strategy to assure the boundedness of the adaptive gains also in presence of persistent disturbances. The closed-loop system is analytically studied, and conditions for the asymptotic convergence of the tracking error are presented. Furthermore, when the plant is subjected to unmatched disturbances, the ultimate boundedness of the closed-loop dynamics, which are made discontinuous by the adaptive switching control actions, is systematically proven by using Lyapunov theory for Filippov systems. The problem of trajectory tracking for space robotic arms in presence of unknown and noncooperative targets is used to test the effectiveness of the novel multi-input EMRAC algorithms for taming uncertain systems. Four EMRAC solutions are designed for this engineering application, and tested within a high fidelity simulation framework based on the Robot Operating System. Finally, the tracking performance of the EMRAC implementations is quantitatively evaluated via a set of key performance indicators in the joint space and operational space, and compared with that of four benchmarking controllers.

To extend planetary exploration beyond the current limitations of wheeled vehicles while preserving reliability, simplicity, and efficiency, actuation can be judiciously incorporated into the locomotion system. Based on a static analysis, we propose a new four-wheeled chassis concept for planetary rovers that can traverse more challenging terrain with the help of two internal active joints. These joints are arranged as follows: a vertical pivot articulates the chassis around its center while a bogie allows the rear wheels to rotate around the longitudinal axis of the vehicle. We also introduce a control method that uses a two-stage procedure to produce an interpretable controller based on a policy devised by reinforcement learning. This way, we eliminate the black box made of a neural network and facilitate the transfer from simulation to reality. The resulting controller efficiently harnesses the internal mobility of the chassis to climb over obstacles in a sequenced manner while relying only on proprioceptive data provided by the chassis. A rover prototype named MARCEL has been built and tested experimentally. Contrary to any state-of-the-art six-wheeled passive chassis, the proposed locomotion system and its associated control has proven to be able to overcome solid step obstacles as tall as the diameter of the wheels with a 90 degrees edge and a friction coefficient as low as 0.5. This simple but capable design will enable future missions to explore more challenging areas while providing better guarantees in the face of unforeseen difficulties that could arise.

Driving energy consumption plays an important role in the navigation of autonomous mobile robots in off-road scenarios. However, the accuracy of the driving energy predictions is often affected by a high degree of uncertainty due to unknown and constantly varying terrain properties, and the complex wheel-terrain interaction in unstructured terrains. In this paper, a probabilistic deep meta-learning approach is proposed to model the existing uncertainty in the driving energy consumption and efficiently adapt the probabilistic predictions based on a small number of local measurements. The method expands upon an existing deterministic deep-meta learning model that, in contrast, only provided single-point energy estimates. The performance of the proposed method is compared against the deterministic approach in a 3D-body dynamic simulator over several typologies of deformable terrains and unstructured geometries. In this way, the benefits of the proposed method are illustrated to enhance the predictions with informative probabilistic considerations, which can be crucial to the safety of mobile robots traversing challenging, unstructured environments.

The Lunar Volatile and Mineralogy Mapping Orbiter (VMMO) comprises a low-cost 12U Cubesat with deployable solar arrays, X-Band/UHF communications, option of electric or chemical propulsion, the Lunar Volatiles and Mineralogy Mapper (LVMM) payload, and an optional GPS receiver technology demonstrator. LVMM facilitates three operational modes: Active mode using illumination of the lunar surface at 532nm, 1064nm, and 1560nm to enable volatiles mapping during the lunar night and within Permanently-Shadowed Regions (PSRs); Passive mode during the lunar day with spectral channels at 300nm, 532nm, 690nm, 1064nm, and 1560nm for mapping lunar surficial ilmenite (FeTiO3); and a Communications mode for an optical data downlink demonstration at 1560nm. Previous lunar missions have detected the presence of water-ice in the lunar South Pole region. However, there is considerable uncertainty with regards to its distribution within and across the lunar surface. A number of planned future missions will further map water ice deposits, but the spatial resolution of these observations is expected to be on the order of kilometres. The LVMM using single-mode fiber lasers can improve the special resolution of the mapping to 10s of metres. VMMO has completed the Phase A study with ESA. This paper discusses the baseline LVMM payload design and its dual-use applications for both the stand-off mapping of lunar volatiles and a high-speed optical data link demonstration. In particular, the supporting fiber-laser technology readiness was advanced through ground qualification.

Traversing soft soils represents a major concern of planetary rover missions. In this paper, we present a new chassis mechanism capable of a crawling gait that enhances trafficability on soft soil while relying on as few actuators as possible. Articulated by two actuated joints, MARCEL is a four-wheeled rover chassis which name stands for Mobile Active Rover Chassis for Enhanced Locomotion. MARCEL's crawling leverages a continuous adjustment of the load distribution on the four wheels using an internal torque applied between two halves of the chassis by series elastic actuation. This allows the pressure on two wheels to be minimized while they are moving forward with the assistance of the chassis's articulated motion. As a result, the wheels can be propelled forward one pair after another while avoiding the bulldozing resistance of the sand. This crawling motion is tested experimentally and is shown to generate more drawbar pull than both rolling or using a mere “push-pull” locomotion. Its ability to extricate the rover from deep sand entrapment is also tested successfully. This will allow future missions to deal with unforeseen terrain properties or to venture in more challenging areas while minimizing design complexity.

Driving energy consumption plays a major role in the navigation of mobile robots in challenging environments, especially if they are left to operate unattended under limited on-board power. This paper reports on first results of an energy-aware path planner, which can provide estimates of the driving energy consumption and energy recovery of a robot traversing complex uneven terrains. Energy is estimated over trajectories making use of a self-supervised learning approach, in which the robot autonomously learns how to correlate perceived terrain point clouds to energy consumption and recovery. A novel feature of the method is the use of 1D convolutional neural network to analyse the terrain sequentially in the same temporal order as it would be experienced by the robot when moving. The performance of the proposed approach is assessed in simulation over several digital terrain models collected from real natural scenarios, and is compared with a heuristic inclination-based energy model. We show evidence of the benefit of our method to increase the overall prediction r2 score by 66.8% and to reduce the driving energy consumption over planned paths by 5.5%.

Active debris removal in space has become a necessary activity to maintain and facilitate orbital operations. Current approaches tend to adopt autonomous robotic systems which are often furnished with a robotic arm to safely capture debris by identifying a suitable grasping point. These systems are controlled by mission-critical software, where a software failure can lead to mission failure which is difficult to recover from since the robotic systems are not easily accessible to humans. Therefore, verifying that these autonomous robotic systems function correctly is crucial. Formal verification methods enable us to analyse the software that is controlling these systems and to provide a proof of correctness that the software obeys its requirements. However, robotic systems tend not to be developed with verification in mind from the outset, which can often complicate the verification of the final algorithms and systems. In this paper, we describe the process that we used to verify a pre-existing system for autonomous grasping which is to be used for active debris removal in space. In particular, we formalise the requirements for this system using the Formal Requirements Elicitation Tool (FRET). We formally model specific software components of the system and formally verify that they adhere to their corresponding requirements using the Dafny program verifier. From the original FRET requirements, we synthesise runtime monitors using ROSMonitoring and show how these can provide runtime assurances for the system. We also describe our experimentation and analysis of the testbed and the associated simulation. We provide a detailed discussion of our approach and describe how the modularity of this particular autonomous system simplified the usually complex task of verifying a system post-development.

Improving the efficiency of a subsurface sample acquisition process provides operational benefits such as reduced bit wear and power consumption, and may increase the sample fidelity by minimising alterations caused by the drilling operation to the surrounding substrate. This could be achieved by acquiring multiple samples during a single drilling procedure. Although some recent planetary drills have incorporated a hybrid cored and cuttings sampling technique, there are currently no systems that have been designed to obtain and separately store multiple comparable samples. The Internal Actuation Mechanism, developed as part of the latest generation of the Dual-Reciprocating Drill, will incorporate a rotating shutter mechanism capable of acquiring up to four samples. In the first demonstration of the fully-integrated prototype using layers of differently-coloured sand, the shutter mechanism was shown to be able to take multiple samples in a single drilling operation. These experiments also confirmed the observations of numerical simulations, which showed that the drill's teethed design results in regolith from the surface layer being dragged down with the descending drill and collected by the sampling system. By optimising the geometrical design to increase sample value and refining the shutter mechanism to improve reliability, a multi-sample acquisition system for planetary subsurface exploration could become a viable technology.