arXiv subjects

Georg Stettinger

Publications and source records attributed to Georg Stettinger.

15 recordsLinked to original sources

arXiv2026

The BEF Symplectic Form: A Lagrangian Perspective

In 2025, Bernardes, Erler and Firat proposed a novel, elegant expression for the symplectic form on phase space applicable to non-local theories. We show that this BEF symplectic structure can be derived directly from an $L_\infty$-Lagrangian by following the covariant phase space approach. Moreover, we establish a precise relation between the BEF symplectic structure and the Barnich--Brandt symplectic form for general finite-derivative theories. In particular, we prove that for theories with second-order equations of motion, the BEF symplectic structure coincides with the Barnich--Brandt construction, thereby explaining the emergence of the canonical corner term in general relativity within the BEF approach. We further argue that the BEF symplectic structure naturally encodes information about generic corner terms and some information about boundary conditions. In addition, we develop a general expression for the Hamiltonian in theories in $L_\infty$-form and present several explicit examples illustrating the construction.

hep-th↗

arXiv2025

XD-RCDepth: Lightweight Radar-Camera Depth Estimation with Explainability-Aligned and Distribution-Aware Distillation

Depth estimation remains central to autonomous driving, and radar-camera fusion offers robustness in adverse conditions by providing complementary geometric cues. In this paper, we present XD-RCDepth, a lightweight architecture that reduces the parameters by 29.7% relative to the state-of-the-art lightweight baseline while maintaining comparable accuracy. To preserve performance under compression and enhance interpretability, we introduce two knowledge-distillation strategies: an explainability-aligned distillation that transfers the teacher's saliency structure to the student, and a depth-distribution distillation that recasts depth regression as soft classification over discretized bins. Together, these components reduce the MAE compared with direct training with 7.97% and deliver competitive accuracy with real-time efficiency on nuScenes and ZJU-4DRadarCam datasets. Code: https://github.com/harborsarah/XD_RCDepth

cs.CV↗

arXiv2025

Selected topics in string field theory

This thesis contains of two parts: The first part is a pedagogical introduction into the field of bosonic SFT. After discussing some general properties we expect, Witten's open SFT and Zwiebach's closed SFT are presented in detail. This means we set up the action, explain the algebraical and geometrical structure and mention physical applications. In the open case we review the most important analytic solutions including the necessary operator technology whileas in the closed case we focus on quantization and the more elaborate construction of the vertices. The second part contains of reprints of papers completed within the PhD: 2301.13182, 2402.00308, 2410.16228 and 2411.15123.

hep-th↗

arXiv2025

What's Really Different with AI? -- A Behavior-based Perspective on System Safety for Automated Driving Systems

Assuring safety for ``AI-based'' systems is one of the current challenges in safety engineering. For automated driving systems, in particular, further assurance challenges result from the open context that the systems need to operate in after deployment. The current standardization and regulation landscape for ``AI-based'' systems is becoming ever more complex, as standards and regulations are being released at high frequencies. This position paper seeks to provide guidance for making qualified arguments which standards should meaningfully be applied to (``AI-based'') automated driving systems. Furthermore, we argue for clearly differentiating sources of risk between AI-specific and general uncertainties related to the open context. In our view, a clear conceptual separation can help to exploit commonalities that can close the gap between system-level and AI-specific safety analyses, while ensuring the required rigor for engineering safe ``AI-based'' systems.

eess.SY↗

arXiv2025

CaRaFFusion: Improving 2D Semantic Segmentation with Camera-Radar Point Cloud Fusion and Zero-Shot Image Inpainting

Segmenting objects in an environment is a crucial task for autonomous driving and robotics, as it enables a better understanding of the surroundings of each agent. Although camera sensors provide rich visual details, they are vulnerable to adverse weather conditions. In contrast, radar sensors remain robust under such conditions, but often produce sparse and noisy data. Therefore, a promising approach is to fuse information from both sensors. In this work, we propose a novel framework to enhance camera-only baselines by integrating a diffusion model into a camera-radar fusion architecture. We leverage radar point features to create pseudo-masks using the Segment-Anything model, treating the projected radar points as point prompts. Additionally, we propose a noise reduction unit to denoise these pseudo-masks, which are further used to generate inpainted images that complete the missing information in the original images. Our method improves the camera-only segmentation baseline by 2.63% in mIoU and enhances our camera-radar fusion architecture by 1.48% in mIoU on the Waterscenes dataset. This demonstrates the effectiveness of our approach for semantic segmentation using camera-radar fusion under adverse weather conditions.

cs.CV↗

arXiv2024

A boundary term for open string field theory

We consider Witten's open string field theory in the presence of a non-trivial boundary of spacetime. For the kinetic term, we derive a Gibbons-Hawking-type contribution that has to be added to the action to guarantee a well-defined variational principle. The derivation is done first in a heuristic way and then confirmed by a path integral based approach using the CFT operator formalism. In the last section we discuss the boundary contributions coming from the cubic vertex, although it is problematic to apply consistent boundary conditions on the string field due to the non-locality of the vertex.

hep-th↗

arXiv2024

Sliver frame stubs in OSFT via auxiliary fields

In this short paper we want to generalize some recent concepts related to stubs in open string field theory. First, we modify the auxiliary field method by Erbin and Firat [#Erbin2023] to non-BPZ even sliver frame stubs. We then also show that the construction is consistent at the full quantum level without any additional assumptions. Finally, we apply the method explicitly to the tachyon vacuum and the simplest identity-like solution.

hep-th↗

arXiv2024

Enhanced Radar Perception via Multi-Task Learning: Towards Refined Data for Sensor Fusion Applications

Radar and camera fusion yields robustness in perception tasks by leveraging the strength of both sensors. The typical extracted radar point cloud is 2D without height information due to insufficient antennas along the elevation axis, which challenges the network performance. This work introduces a learning-based approach to infer the height of radar points associated with 3D objects. A novel robust regression loss is introduced to address the sparse target challenge. In addition, a multi-task training strategy is employed, emphasizing important features. The average radar absolute height error decreases from 1.69 to 0.25 meters compared to the state-of-the-art height extension method. The estimated target height values are used to preprocess and enrich radar data for downstream perception tasks. Integrating this refined radar information further enhances the performance of existing radar camera fusion models for object detection and depth estimation tasks.

cs.CV↗

arXiv2024

More on stubs in open string field theory

We continue our analysis of open string field theory based on A-infinity-algebras obtained from Witten's theory by attaching stubs to the elementary vertex. Classical solutions of the new theory can be obtained from known analytic solutions in Witten's theory by applying a cohomomorphism. In a previous work, two such cohomomorphisms were found, one non-cyclic, obtained from the homological perturbation lemma and another cyclic one by geometric methods. Here we show that the two resulting maps are related by a combination of a gauge transformation and a term vanishing on-shell. In the second part of the paper we generalize the whole construction to non-BPZ even stubs, in particular sliver frame stubs. We discuss algebraic and geometric aspects and analyze the resulting conditions on the homotopy operator. Moreover, we explicitly calculate the first few orders of the new A-infinity-algebra solution for the tachyon vacuum in the sliver frame.

hep-th↗

arXiv2023

Multi-Task Cross-Modality Attention-Fusion for 2D Object Detection

Accurate and robust object detection is critical for autonomous driving. Image-based detectors face difficulties caused by low visibility in adverse weather conditions. Thus, radar-camera fusion is of particular interest but presents challenges in optimally fusing heterogeneous data sources. To approach this issue, we propose two new radar preprocessing techniques to better align radar and camera data. In addition, we introduce a Multi-Task Cross-Modality Attention-Fusion Network (MCAF-Net) for object detection, which includes two new fusion blocks. These allow for exploiting information from the feature maps more comprehensively. The proposed algorithm jointly detects objects and segments free space, which guides the model to focus on the more relevant part of the scene, namely, the occupied space. Our approach outperforms current state-of-the-art radar-camera fusion-based object detectors in the nuScenes dataset and achieves more robust results in adverse weather conditions and nighttime scenarios.

cs.CV↗

arXiv2023

Open string field theory with stubs

There are various reasons why adding stubs to the vertices of open string field theory (OSFT) is interesting: Not only the stubs can tame certain singularities and make the theory more well-behaved, but also the new theory shares a lot of similarities with closed string field theory, which helps to improve our understanding of its structure and possible solutions. In this paper we explore two natural ways of implementing stubs into the framework of OSFT, resulting in an A-infinity-algebra giving rise to infinitely many vertices. We find two distinct consistent actions, both generated by a field redefinition, interestingly sharing the same equations of motion. In the last section we illustrate their relationship and physical meaning by applying our construction to nearly marginal solutions.

hep-th↗

arXiv2020

Overview of Tools Supporting Planning for Automated Driving

Planning is an essential topic in the realm of automated driving. Besides planning algorithms that are widely covered in the literature, planning requires different software tools for its development, validation, and execution. This paper presents a survey of such tools including map representations, communication, traffic rules, open-source planning stacks and middleware, simulation, and visualization tools as well as benchmarks. We start by defining the planning task and different supporting tools. Next, we provide a comprehensive review of state-of-the-art developments and analysis of relations among them. Finally, we discuss the current gaps and suggest future research directions.

cs.RO↗

arXiv2019

Search-Based Motion Planning for Performance Autonomous Driving

Driving on the limits of vehicle dynamics requires predictive planning of future vehicle states. In this work, a search-based motion planning is used to generate suitable reference trajectories of dynamic vehicle states with the goal to achieve the minimum lap time on slippery roads. The search-based approach enables to explicitly consider a nonlinear vehicle dynamics model as well as constraints on states and inputs so that even challenging scenarios can be achieved in a safe and optimal way. The algorithm performance is evaluated in simulated driving on a track with segments of different curvatures.

cs.RO↗

arXiv2018

New massless and massive infinite derivative gravity in three dimensions

In this paper we will consider the most general quadratic curvature action with infinitely many covariant derivatives of massless gravity in three spacetime dimensions. The action is parity invariant and torsion-free and contains the same off-shell degrees of freedom as the Einstein-Hilbert action in general relativity. In the ultraviolet, with an appropriate choice of the propagator given by the exponential of an entire function, the point-like curvature singularity can be smoothened to a Gaussian distribution, while in the infrared the theory reduces to general relativity. We will also show how to embed new massive gravity in ghost-free infinite derivative gravity in Minkowski background as one of the infrared limits. Finally, we will provide the tree-level unitarity conditions for infinite derivative gravity in presence of a cosmological constant in deSitter and Anti-deSitter spacetimes in three dimensions by perturbing the geometries.

hep-th↗

arXiv2018

Safe learning-based optimal motion planning for automated driving

This paper presents preliminary work on learning the search heuristic for the optimal motion planning for automated driving in urban traffic. Previous work considered search-based optimal motion planning framework (SBOMP) that utilized numerical or model-based heuristics that did not consider dynamic obstacles. Optimal solution was still guaranteed since dynamic obstacles can only increase the cost. However, significant variations in the search efficiency are observed depending whether dynamic obstacles are present or not. This paper introduces machine learning (ML) based heuristic that takes into account dynamic obstacles, thus adding to the performance consistency for achieving real-time implementation.

cs.LG↗