Searcharxiv⌕ Search

arXiv subjects

Sebastian Stock

Publications and source records attributed to Sebastian Stock.

17 recordsLinked to original sources
arXiv2026

Agentic AI for Robot Control: Flexible but still Fragile

Recent work leverages the capabilities and commonsense priors of generative models for robot control. In this paper, we present an agentic control system in which a reasoning-capable language model plans and executes tasks by selecting and invoking robot skills within an iterative planner and executor loop. We deploy the system on two physical robot platforms in two settings: (i) tabletop grasping, placement, and box insertion in indoor mobile manipulation (Mobipick) and (ii) autonomous agricultural navigation and sensing (Valdemar). Both settings involve uncertainty, partial observability, sensor noise, and ambiguous natural-language commands. The system exposes structured introspection of its planning and decision process, reacts to exogenous events via explicit event checks, and supports operator interventions that modify or redirect ongoing execution. Across both platforms, our proof-of-concept experiments reveal substantial fragility, including non-deterministic suboptimal behavior, instruction-following errors, and high sensitivity to prompt specification. At the same time, the architecture is flexible: transfer to a different robot and task domain largely required updating the system prompt (domain model, affordances, and action catalogue) and re-binding the same tool interface to the platform-specific skill API.

cs.RO↗
arXiv2025

Core-Shell Confinement Blocks Hydride Formation: The Impact of Surface Oxides on Hydrogen Sorption in Nanoporous FeTi

Metal hydrides remain an intriguing alternative to conventional gaseous and liquid hydrogen storage methods, offering high volumetric storage density and enhanced hydrogen storage safety at ambient conditions. In this regard, the intermetallic compound FeTi is one of the most promising storage materials. However, its widespread industrial application remains challenging due to the need for activation, slow initial kinetics, large hysteresis, and high material costs. In this study, we aim to overcome these limitations by devising an alternative synthesis pathway to prepare nanoporous and ultra-fine porous FeTi with controlled grain and ligament sizes, allowing us to study the obtained well-defined microstructures in detail. In particular, we observe the confinement of the FeTi phase by surface oxides, which can be correlated with the hydrogen sorption properties of the respective material. These experimental results are further supported by an analytical model allowing the calculation of the absorption pressure as a function of microstructure-dependent elastic stresses. Additionally, we show that such stresses also influence the absorption-desorption hysteresis. This study lays the groundwork for the controlled and systematic study of the processing-structure-properties relations in metal hydrides and FeTi in particular, thereby paving the way to cost-effective and efficient hydrogen storage solutions based on metal hydrides.

cond-mat.mtrl-sci↗
arXiv2025

Uncertainty-Resilient Active Intention Recognition for Robotic Assistants

Purposeful behavior in robotic assistants requires the integration of multiple components and technological advances. Often, the problem is reduced to recognizing explicit prompts, which limits autonomy, or is oversimplified through assumptions such as near-perfect information. We argue that a critical gap remains unaddressed -- specifically, the challenge of reasoning about the uncertain outcomes and perception errors inherent to human intention recognition. In response, we present a framework designed to be resilient to uncertainty and sensor noise, integrating real-time sensor data with a combination of planners. Centered around an intention-recognition POMDP, our approach addresses cooperative planning and acting under uncertainty. Our integrated framework has been successfully tested on a physical robot with promising results.

cs.RO↗
arXiv2025

Acting and Planning with Hierarchical Operational Models on a Mobile Robot: A Study with RAE+UPOM

Robotic task execution faces challenges due to the inconsistency between symbolic planner models and the rich control structures actually running on the robot. In this paper, we present the first physical deployment of an integrated actor-planner system that shares hierarchical operational models for both acting and planning, interleaving the Reactive Acting Engine (RAE) with an anytime UCT-like Monte Carlo planner (UPOM). We implement RAE+UPOM on a mobile manipulator in a real-world deployment for an object collection task. Our experiments demonstrate robust task execution under action failures and sensor noise, and provide empirical insights into the interleaved acting-and-planning decision making process.

cs.RO↗
arXiv2025

Failure divergence refinement for Event-B

When validating formal models, sizable effort goes into ensuring two types of properties: safety properties (nothing bad happens) and liveness properties (something good occurs eventually. Event-B supports checking safety properties all through the refinement chain. The same is not valid for liveness properties. Liveness properties are commonly validated with additional techniques like animation, and results do not transfer quickly, leading to re-doing the validation process at every refinement stage. This paper promotes early validation by providing failure divergence refinement semantics for Event-B. We show that failure divergence refinement preserves trace properties, which comprise many liveness properties, under certain natural conditions. Consequently, re-validation of those properties becomes unnecessary. Our result benefits data refinements, where no abstract behavior should be removed during refinement. Furthermore, we lay out an algorithm and provide a tool for automatic failure divergence refinement checking, significantly decreasing the modeler's workload. The tool is compared and evaluated in the context of sizable case studies.

cs.LO↗
arXiv2025

In situ vs ex situ: Comparing the structure of PNIPAM microgels at the air/water and air/solid interfaces

For studying the structure of microgel particles at the air/water interface, specular and off-specular X-ray reflectivity (OSR/XRR) allows in situ measurements without any labelling techniques. Herein we investigate the vertical and lateral structure of poly(N-isopropylacrylamide) (PNIPAM) microgels (MGs) at the air/water interface and the effect of Langmuir-Blodgett (LB) transfer onto solid substrates. The initial ex situ atomic force microscopy (AFM) scans of LB-transferred MGs at the air/solid interface reveal strong lateral 2D hexagonal ordering across a broad range of lateral surface pressures at the air/water interface before LB-transfer. Notably, for the first time, these results were confirmed by OSR, demonstrating the existence of the long-range hexagonal ordering at low and intermediate surface pressures. For in situ conditions and upon uniaxial compression at the air/water interface, the MG lattice constant decreases non-monotonically. This indicates the formation of domains at low pressures that approach each other and only compress when the surface isotherm reaches a plateau. Comparing the results of in situ and ex situ measurements, our study demonstrates a clear transfer effect during the LB-deposition on the lateral ordering of the MGs: the distance between the particles decreased during LB-transfer, and at high pressures ($\Pi\,>\,17\,\mathrm{mNm^{-1}}$) a second distance occurs indicating small domains with hexagonal internal ordering. The novel surface characterisation approaches debuted here highlight the use of both XRR and OSR to probe the vertical and lateral structure of adsorbed MGs, offering in situ, non-invasive insights without the need for doping or transfer-induced artefacts.

cond-mat.soft↗
arXiv2024

Application of AI to formal methods - an analysis of current trends

Context: With artificial intelligence (AI) being well established within the daily lives of research communities, we turn our gaze toward formal methods (FM). FM aim to provide sound and verifiable reasoning about problems in computer science. Objective: We conduct a systematic mapping study to overview the current landscape of research publications that apply AI to FM. We aim to identify how FM can benefit from AI techniques and highlight areas for further research. Our focus lies on the previous five years (2019-2023) of research. Method: Following the proposed guidelines for systematic mapping studies, we searched for relevant publications in four major databases, defined inclusion and exclusion criteria, and applied extensive snowballing to uncover potential additional sources. Results: This investigation results in 189 entries which we explored to find current trends and highlight research gaps. We find a strong focus on AI in the area of theorem proving while other subfields of FM are less represented. Conclusions: The mapping study provides a quantitative overview of the modern state of AI application in FM. The current trend of the field is yet to mature. Many primary studies focus on practical application, yet we identify a lack of theoretical groundwork, standard benchmarks, or case studies. Further, we identify issues regarding shared training data sets and standard benchmarks.

cs.LO↗
arXiv2023

Conceptualizing flexible papers using cellulose model surfaces and polymer particles

Cellulose, as a naturally abundant and biocompatible material, is still gaining interest due to its high potential for functionalization. This makes cellulose a promising candidate for replacing plastics. Understanding how cellulose interacts with various additives is crucial for creating composite materials with diverse properties, as it is the case for plastics. In addition, the mechanical properties of the composite materials are assumed to be related to the mobility of the additives against the cellulose. Using a well-defined cellulose model surface (CMS), we aim to understand the adsorption and desorption of two polymeric particles (core-shell particles and microgels) to/from the cellulose surface. The nanomechanics of particles and CMS are quantified by indentation measurements with an atomic force microscope (AFM). AFM topography measurements quantified particle adsorption and desorption on the CMS, while peak force AFM measurements determined the force needed to move individual particles. Both particles and the CMS exhibited pH-dependent charge behavior, allowing a tunable interaction between them. Particle adsorption was irreversible and driven by electrostatic forces. In contrast, desorption and particle mobility forces are dominated by structural morphology. In addition, we found that an annealing procedure consisting of swelling/drying cycles significantly increased the adhesion strength of both particles. Using the data, we achieve a deeper understanding of the interaction of cellulose with polymeric particles, with the potential to advance the development of functional materials and contribute to various fields, including smart packaging, sensors, and biomedical applications.

cond-mat.soft↗
arXiv2023

Validation-Driven Development

Formal methods play a fundamental role in asserting the correctness of requirements specifications. However, historically, formal method experts have primarily focused on verifying those specifications. Although equally important, validation of requirements specifications often takes the back seat. This paper introduces a validation-driven development (VDD) process that prioritizes validating requirements in formal development. The VDD process is built upon problem frames - a requirements analysis approach - and validation obligations (VOs) - the concept of breaking down the overall validation of a specification and linking it to refinement steps. The effectiveness of the VDD process is demonstrated through a case study in the aviation industry.

cs.SE↗
arXiv2023

Impact of Ultrasound on the Motion of Compact Particles and Acousto-responsive Microgels

In this study, we investigate dynamic light scattering (DLS) from both randomly diffusing silica particles and acousto-responsive microgels in aqueous dispersions under ultrasonic vibration. Employing high-frequency ultrasound (US) with low amplitude ensures that the polymers remain intact without damage. We derive theoretical expressions for the homodyne autocorrelation function, incorporating the US term alongside the diffusion term. Subsequently, we successfully combine US with a conventional DLS system to experimentally characterize compact silica particles and microgels under the influence of US. Our model allows us to extract essential parameters, including particle size, frequency, and amplitude of particle vibration, based on the correlation function of the scattered light intensity. The studies involving non-responsive silica particles demonstrate that US does not disrupt size determination, establishing them as suitable reference systems. Microgels show the same swelling/shrinking behavior as that induced by temperature, but with significantly faster kinetics. The findings of this study have potential applications in various industrial and biomedical fields that benefit from the characterization of macromolecules subjected to US.

cond-mat.soft↗
arXiv2022

Trace Refinement in B and Event-B

Traces are used to show whether a model complies with the intended behavior. A modeler can use trace checking to ensure the preservation of the model behavior during the refinement process. In this paper, we present a trace refinement technique and tool called BERT that allows designers to ensure the behavioral integrity of high-level traces at the concrete level. The proposed technique is evaluated within the context of the B and Event-B methods on industrial-strength case studies from the automotive domain.

cs.LO↗
arXiv2022

Application of Validation Obligations to Security Concerns

Our lives become increasingly dependent on safety- and security-critical systems, so formal techniques are advocated for engineering such systems. One of such techniques is validation obligations that enable formalizing requirements early in development to ensure their correctness. Furthermore, validation obligations help hold requirements consistent in an evolving model and create assurances about the model's completeness. Although initially proposed for safety properties, this paper shows how the technique of validation obligations enables us to also reason about security concerns through an example from the medical domain.

cs.LO↗
arXiv2010

Gauge Deformations and Embedding Theorems for Special Geometries

We reduce the embedding problem for hypo SU(2) and SU(3)-structures to the embedding problem for hypo G2-structures into parallel Spin(7)-manifolds. The latter will be described in terms of gauge deformations. This description involves the intrinsic torsion of the initial G2-structure and allows us to prove that the evolution equations, for all of the above embedding problems, do not admit non-trivial longtime solutions.

math.DG↗
arXiv2007

Lifting SU(3)-structures to nearly parallel G_{2}-structures

Hitchin shows that half-flat SU(3)-structures on a 6-dimensional manifold M can be lifted to parallel G_{2}-structure on the product $M\times\mathbb{R}$. We show that Hitchin's approach can also be used to construct nearly parallel G_{2}-structures by lifting so-called nearly half-flat structures. These SU(3)-structures are described by pairs $(ω,ϕ)$ of stable 2- and 3-forms with $dϕ=λω^{2}$ for some $λ\in\mathbb{R}\setminus\{0\}$.

math.DG↗