Hanno Gottschalk

Manuel Schwonberg, Joshua Niemeijer, Jan-Aike Termöhlen et al.

Deep neural networks (DNNs) have proven their capabilities in many areas in the past years, such as robotics, or automated driving, enabling technological breakthroughs. DNNs play a significant role in environment perception for the challenging application of automated driving and are employed for tasks such as detection, semantic segmentation, and sensor fusion. Despite this progress and tremendous research efforts, several issues still need to be addressed that limit the applicability of DNNs in automated driving. The bad generalization of DNNs to new, unseen domains is a major problem on the way to a safe, large-scale application, because manual annotation of new domains is costly, particularly for semantic segmentation. For this reason, methods are required to adapt DNNs to new domains without labeling effort. The task, which these methods aim to solve is termed unsupervised domain adaptation (UDA). While several different domain shifts can challenge DNNs, the shift between synthetic and real data is of particular importance for automated driving, as it allows the use of simulation environments for DNN training. In this work, we present an overview of the current state of the art in this field of research. We categorize and explain the different approaches for UDA. The number of considered publications is larger than any other survey on this topic. The scope of this survey goes far beyond the description of the UDA state-of-the-art. Based on our large data and knowledge base, we present a quantitative comparison of the approaches and use the observations to point out the latest trends in this field. In the following, we conduct a critical analysis of the state-of-the-art and highlight promising future research directions. With this survey, we aim to facilitate UDA research further and encourage scientists to exploit novel research directions to generalize DNNs better.

Sebastian Schmitz, Georg Rollmann, Hanno Gottschalk et al.

An accurate risk assessment for fatigue damage is of vital importance for the design and service of today's turbomachinery components. We present an approach for quantifying the probability of crack initiation due to surface driven low-cycle fatigue (LCF). This approach is based on the theory of failure-time processes and takes inhomogeneous stress fields and size effects into account. The method has been implemented as a finite-element postprocessor which uses quadrature formulae of higher order. Results of applying this new approach to an example case of a gas-turbine compressor disk are discussed.

Robin Chan, Radin Dardashti, Meike Osinski et al.

Deep neural networks (DNN) have made impressive progress in the interpretation of image data, so that it is conceivable and to some degree realistic to use them in safety critical applications like automated driving. From an ethical standpoint, the AI algorithm should take into account the vulnerability of objects or subjects on the street that ranges from "not at all", e.g. the road itself, to "high vulnerability" of pedestrians. One way to take this into account is to define the cost of confusion of one semantic category with another and use cost-based decision rules for the interpretation of probabilities, which are the output of DNNs. However, it is an open problem how to define the cost structure, who should be in charge to do that, and thereby define what AI-algorithms will actually "see". As one possible answer, we follow a participatory approach and set up an online survey to ask the public to define the cost structure. We present the survey design and the data acquired along with an evaluation that also distinguishes between perspective (car passenger vs. external traffic participant) and gender. Using simulation based $F$-tests, we find highly significant differences between the groups. These differences have consequences on the reliable detection of pedestrians in a safety critical distance to the self-driving car. We discuss the ethical problems that are related to this approach and also discuss the problems emerging from human-machine interaction through the survey from a psychological point of view. Finally, we include comments from industry leaders in the field of AI safety on the applicability of survey based elements in the design of AI functionalities in automated driving.

Hanno Gottschalk, Mohamed Saadi, Onur Tanil Doganay et al.

In the optimization of turbomachinery components, shape sensitivities for fluid dynamical objective functions have been used for a long time. As peak stress is not a differential func- tional of the shape, such highly efficient procedures so far have been missing for objective functionals that stem from mechan- ical integrity. This changes, if deterministic lifing criteria are replaced by probabilistic criteria, which have been introduced recently to the field of low cycle fatigue (LCF). Here we present a finite element (FEA) based first discretize, then adjoin approach to the calculation of shape gradients (sen- sitivities) for the failure probability with regard to probabilistic LCF and apply it to simple and complex geometries, as e.g. a blisk geometry. We review the computation of failure probabilities with a FEA postprocessor and sketch the computation of the relevant quantities for the adjoint method. We demonstrate high accuracy and computational efficiency of the adjoint method compared to finite difference schemes. We discuss implementation details for rotating components with cyclic boundary conditions. Finally, we shortly comment on future development steps and on poten- tial applications in multi criteria optimization.

Lucas Mäde, Sebastian Schmitz, Georg Rollmann et al.

A probabilistic risk assessment for low cycle fatigue (LCF) based on the so-called size effect has been applied on gas-turbine design in recent years. In contrast, notch support modeling for LCF which intends to consider the change in stress below the surface of critical LCF regions is known and applied for decades. Turbomachinery components often show sharp stress gradients and very localized critical regions for LCF crack initiations so that a life prediction should also consider notch and size effects. The basic concept of a combined probabilistic model that includes both, size effect and notch support, is presented. In many cases it can improve LCF life predictions significantly, in particular compared to \textit{E-N} curve predictions of standard specimens where no notch support and size effect is considered. Here, an application of such a combined model is shown for a turbine vane.

Kira Maag, Robin Chan, Svenja Uhlemeyer et al.

In this work we present two video test data sets for the novel computer vision (CV) task of out of distribution tracking (OOD tracking). Here, OOD objects are understood as objects with a semantic class outside the semantic space of an underlying image segmentation algorithm, or an instance within the semantic space which however looks decisively different from the instances contained in the training data. OOD objects occurring on video sequences should be detected on single frames as early as possible and tracked over their time of appearance as long as possible. During the time of appearance, they should be segmented as precisely as possible. We present the SOS data set containing 20 video sequences of street scenes and more than 1000 labeled frames with up to two OOD objects. We furthermore publish the synthetic CARLA-WildLife data set that consists of 26 video sequences containing up to four OOD objects on a single frame. We propose metrics to measure the success of OOD tracking and develop a baseline algorithm that efficiently tracks the OOD objects. As an application that benefits from OOD tracking, we retrieve OOD sequences from unlabeled videos of street scenes containing OOD objects.

Manuel Schwonberg, Fadoua El Bouazati, Nico M. Schmidt et al.

Unsupervised Domain Adaptation (UDA) and domain generalization (DG) are two research areas that aim to tackle the lack of generalization of Deep Neural Networks (DNNs) towards unseen domains. While UDA methods have access to unlabeled target images, domain generalization does not involve any target data and only learns generalized features from a source domain. Image-style randomization or augmentation is a popular approach to improve network generalization without access to the target domain. Complex methods are often proposed that disregard the potential of simple image augmentations for out-of-domain generalization. For this reason, we systematically study the in- and out-of-domain generalization capabilities of simple, rule-based image augmentations like blur, noise, color jitter and many more. Based on a full factorial design of experiment design we provide a systematic statistical evaluation of augmentations and their interactions. Our analysis provides both, expected and unexpected, outcomes. Expected, because our experiments confirm the common scientific standard that combination of multiple different augmentations out-performs single augmentations. Unexpected, because combined augmentations perform competitive to state-of-the-art domain generalization approaches, while being significantly simpler and without training overhead. On the challenging synthetic-to-real domain shift between Synthia and Cityscapes we reach 39.5% mIoU compared to 40.9% mIoU of the best previous work. When additionally employing the recent vision transformer architecture DAFormer we outperform these benchmarks with a performance of 44.2% mIoU

Julian Burghoff, Matthias Rottmann, Jill von Conta et al.

In this work, we develop a neural architecture search algorithm, termed Resbuilder, that develops ResNet architectures from scratch that achieve high accuracy at moderate computational cost. It can also be used to modify existing architectures and has the capability to remove and insert ResNet blocks, in this way searching for suitable architectures in the space of ResNet architectures. In our experiments on different image classification datasets, Resbuilder achieves close to state-of-the-art performance while saving computational cost compared to off-the-shelf ResNets. Noteworthy, we once tune the parameters on CIFAR10 which yields a suitable default choice for all other datasets. We demonstrate that this property generalizes even to industrial applications by applying our method with default parameters on a proprietary fraud detection dataset.

Claudia Drygala, Francesca di Mare, Hanno Gottschalk

Turbulent flow consists of structures with a wide range of spatial and temporal scales which are hard to resolve numerically. Classical numerical methods as the Large Eddy Simulation (LES) are able to capture fine details of turbulent structures but come at high computational cost. Applying generative adversarial networks (GAN) for the synthetic modeling of turbulence is a mathematically well-founded approach to overcome this issue. In this work, we investigate the generalization capabilites of GAN-based synthetic turbulence generators when geometrical changes occur in the flow configuration (e.g. aerodynamic geometric optimization of structures such as airfoils). As training data, we use the flow around a low-pressure turbine (LPT) stator with periodic wake impact obtained from highly resolved LES. To simulate the flow around a LPT stator, we use the conditional deep convolutional GAN framework pix2pixHD conditioned on the position of a rotating wake in front of the stator. For the generalization experiments we exclude images of wake positions located at certain regions from the training data and use the unseen data for testing. We show the abilities and limits of generalization for the conditional GAN by extending the regions of the extracted wake positions successively. Finally, we evaluate the statistical properties of the synthesized flow field by comparison with the corresponding LES results.

Robin Chan, Sarina Penquitt, Hanno Gottschalk

LU-Net is a simple and fast architecture for invertible neural networks (INN) that is based on the factorization of quadratic weight matrices $\mathsf{A=LU}$, where $\mathsf{L}$ is a lower triangular matrix with ones on the diagonal and $\mathsf{U}$ an upper triangular matrix. Instead of learning a fully occupied matrix $\mathsf{A}$, we learn $\mathsf{L}$ and $\mathsf{U}$ separately. If combined with an invertible activation function, such layers can easily be inverted whenever the diagonal entries of $\mathsf{U}$ are different from zero. Also, the computation of the determinant of the Jacobian matrix of such layers is cheap. Consequently, the LU architecture allows for cheap computation of the likelihood via the change of variables formula and can be trained according to the maximum likelihood principle. In our numerical experiments, we test the LU-net architecture as generative model on several academic datasets. We also provide a detailed comparison with conventional invertible neural networks in terms of performance, training as well as run time.

Annika Mütze, Matthias Rottmann, Hanno Gottschalk

Domain adaptation is of huge interest as labeling is an expensive and error-prone task, especially when labels are needed on pixel-level like in semantic segmentation. Therefore, one would like to be able to train neural networks on synthetic domains, where data is abundant and labels are precise. However, these models often perform poorly on out-of-domain images. To mitigate the shift in the input, image-to-image approaches can be used. Nevertheless, standard image-to-image approaches that bridge the domain of deployment with the synthetic training domain do not focus on the downstream task but only on the visual inspection level. We therefore propose a "task aware" version of a GAN in an image-to-image domain adaptation approach. With the help of a small amount of labeled ground truth data, we guide the image-to-image translation to a more suitable input image for a semantic segmentation network trained on synthetic data (synthetic-domain expert). The main contributions of this work are 1) a modular semi-supervised domain adaptation method for semantic segmentation by training a downstream task aware CycleGAN while refraining from adapting the synthetic semantic segmentation expert 2) the demonstration that the method is applicable to complex domain adaptation tasks and 3) a less biased domain gap analysis by using from scratch networks. We evaluate our method on a classification task as well as on semantic segmentation. Our experiments demonstrate that our method outperforms CycleGAN - a standard image-to-image approach - by 7 percent points in accuracy in a classification task using only 70 (10%) ground truth images. For semantic segmentation we can show an improvement of about 4 to 7 percent points in mean Intersection over union on the Cityscapes evaluation dataset with only 14 ground truth images during training.

Tobias Riedlinger, Marius Schubert, Karsten Kahl et al.

Active learning as a paradigm in deep learning is especially important in applications involving intricate perception tasks such as object detection where labels are difficult and expensive to acquire. Development of active learning methods in such fields is highly computationally expensive and time consuming which obstructs the progression of research and leads to a lack of comparability between methods. In this work, we propose and investigate a sandbox setup for rapid development and transparent evaluation of active learning in deep object detection. Our experiments with commonly used configurations of datasets and detection architectures found in the literature show that results obtained in our sandbox environment are representative of results on standard configurations. The total compute time to obtain results and assess the learning behavior can thereby be reduced by factors of up to 14 when comparing with Pascal VOC and up to 32 when comparing with BDD100k. This allows for testing and evaluating data acquisition and labeling strategies in under half a day and contributes to the transparency and development speed in the field of active learning for object detection.

Patrick Krüger, Hanno Gottschalk

Convolutional neural networks revolutionized computer vision and natrual language processing. Their efficiency, as compared to fully connected neural networks, has its origin in the architecture, where convolutions reflect the translation invariance in space and time in pattern or speech recognition tasks. Recently, Cohen and Welling have put this in the broader perspective of invariance under symmetry groups, which leads to the concept of group equivaiant neural networks and more generally steerable neural networks. In this article, we review the architecture of such networks including equivariant layers and filter banks, activation with capsules and group pooling. We apply this formalism to the symmetric group, for which we work out a number of details on representations and capsules that are not found in the literature.

Hanno Gottschalk, Mohamed Saadi

This work provides a numerical calculation of shape gradients of failure probabilities for mechanical components using a first discretize, then adjoint approach. While deterministic life prediction models for failure mechanisms are not (shape) differentiable, this changes in the case of probabilistic life prediction. The probabilistic, or reliability based, approach thus opens the way for efficient adjoint methods in the design for mechanical integrity. In this work we propose, implement and validate a method for the numerical calculation of the shape gradients of failure probabilities for the failure mechanism low cycle fatigue (LCF), which applies to polycrystalline metal. Numerical examples range from a bended rod to a complex geometry from a turbo charger in 3D.

Youssef Shoeb, Robin Chan, Gesina Schwalbe et al.

In the life cycle of highly automated systems operating in an open and dynamic environment, the ability to adjust to emerging challenges is crucial. For systems integrating data-driven AI-based components, rapid responses to deployment issues require fast access to related data for testing and reconfiguration. In the context of automated driving, this especially applies to road obstacles that were not included in the training data, commonly referred to as out-of-distribution (OoD) road obstacles. Given the availability of large uncurated recordings of driving scenes, a pragmatic approach is to query a database to retrieve similar scenarios featuring the same safety concerns due to OoD road obstacles. In this work, we extend beyond identifying OoD road obstacles in video streams and offer a comprehensive approach to extract sequences of OoD road obstacles using text queries, thereby proposing a way of curating a collection of OoD data for subsequent analysis. Our proposed method leverages the recent advances in OoD segmentation and multi-modal foundation models to identify and efficiently extract safety-relevant scenes from unlabeled videos. We present a first approach for the novel task of text-based OoD object retrieval, which addresses the question ''Have we ever encountered this before?''.

Julian Burghoff, Robin Chan, Hanno Gottschalk et al.

Bringing deep neural networks (DNNs) into safety critical applications such as automated driving, medical imaging and finance, requires a thorough treatment of the model's uncertainties. Training deep neural networks is already resource demanding and so is also their uncertainty quantification. In this overview article, we survey methods that we developed to teach DNNs to be uncertain when they encounter new object classes. Additionally, we present training methods to learn from only a few labels with help of uncertainty quantification. Note that this is typically paid with a massive overhead in computation of an order of magnitude and more compared to ordinary network training. Finally, we survey our work on neural architecture search which is also an order of magnitude more resource demanding then ordinary network training.

Svenja Uhlemeyer, Julian Lienen, Eyke Hüllermeier et al.

For open world applications, deep neural networks (DNNs) need to be aware of previously unseen data and adaptable to evolving environments. Furthermore, it is desirable to detect and learn novel classes which are not included in the DNNs underlying set of semantic classes in an unsupervised fashion. The method proposed in this article builds upon anomaly detection to retrieve out-of-distribution (OoD) data as candidates for new classes. We thereafter extend the DNN by $k$ empty classes and fine-tune it on the OoD data samples. To this end, we introduce two loss functions, which 1) entice the DNN to assign OoD samples to the empty classes and 2) to minimize the inner-class feature distances between them. Thus, instead of ground truth which contains labels for the different novel classes, the DNN obtains a single OoD label together with a distance matrix, which is computed in advance. We perform several experiments for image classification and semantic segmentation, which demonstrate that a DNN can extend its own semantic space by multiple classes without having access to ground truth.

Julian Burghoff, Marc Heinrich Monells, Hanno Gottschalk

The highly structured energy landscape of the loss as a function of parameters for deep neural networks makes it necessary to use sophisticated optimization strategies in order to discover (local) minima that guarantee reasonable performance. Overcoming less suitable local minima is an important prerequisite and often momentum methods are employed to achieve this. As in other non local optimization procedures, this however creates the necessity to balance between exploration and exploitation. In this work, we suggest an event based control mechanism for switching from exploration to exploitation based on reaching a predefined reduction of the loss function. As we give the momentum method a port Hamiltonian interpretation, we apply the 'heavy ball with friction' interpretation and trigger breaking (or friction) when achieving certain goals. We benchmark our method against standard stochastic gradient descent and provide experimental evidence for improved performance of deep neural networks when our strategy is applied.

Joris C. Kühl, Hanno Gottschalk

Physical AI is being successfully applied to data which does not follow the traditional paradigm of independent and identically distributed (i.i.d.) samples. In fact, physical AI is often trained on data which is not random at all, and is instead derived from chaotic dynamical systems like turbulence. We aim to explain the empirical success of these methods using the example of generative adversarial networks (GANs), whose statistical learning theory under the i.i.d. assumption is generally well understood. We prove that it is possible, using an infinite-dimensional model of generative adversarial learning (GAL), to learn the invariant distribution of a sufficiently chaotic dynamical system from a single deterministically evolving time series of its states or measurements thereof, and give explicit rates for the convergence to the solution in terms of the Jensen-Shannon divergence.

Laura Fieback, Bidya Dash, Jakob Spiegelberg et al.

Quantifying predictive uncertainty of deep semantic segmentation networks is essential in safety-critical tasks. In applications like autonomous driving, where video data is available, convolutional long short-term memory networks are capable of not only providing semantic segmentations but also predicting the segmentations of the next timesteps. These models use cell states to broadcast information from previous data by taking a time series of inputs to predict one or even further steps into the future. We present a temporal postprocessing method which estimates the prediction performance of convolutional long short-term memory networks by either predicting the intersection over union of predicted and ground truth segments or classifying between intersection over union being equal to zero or greater than zero. To this end, we create temporal cell state-based input metrics per segment and investigate different models for the estimation of the predictive quality based on these metrics. We further study the influence of the number of considered cell states for the proposed metrics.

Patrick Krüger, Patrick Materne, Werner Krebs et al.

Generative learning generates high dimensional data based on low dimensional conditions, also called prompts. Therefore, generative learning algorithms are eligible for solving (Bayesian) inverse problems. In this article we compare a traditional Bayesian inverse approach based on a forward regression model and a prior sampled with the Markov Chain Monte Carlo method with three state of the art generative learning models, namely conditional Generative Adversarial Networks, Invertible Neural Networks and Conditional Flow Matching. We apply them to a problem of gas turbine combustor design where we map six independent design parameters to three performance labels. We propose several metrics for the evaluation of this inverse design approaches and measure the accuracy of the labels of the generated designs along with the diversity. We also study the performance as a function of the training dataset size. Our benchmark has a clear winner, as Conditional Flow Matching consistently outperforms all competing approaches.

Patrick Kruger, Rafael Diaz, Simon Hauschulz et al.

In this paper, we explore the use of generative artificial intelligence (GenAI) for ship propeller design. While traditional forward machine learning models predict the performance of mechanical components based on given design parameters, GenAI models aim to generate designs that achieve specified performance targets. In particular, we employ conditional flow matching to establish a bidirectional mapping between design parameters and simulated noise that is conditioned on performance labels. This approach enables the generation of multiple valid designs corresponding to the same performance targets by sampling over the noise vector. To support model training, we generate data using a vortex lattice method for numerical simulation and analyze the trade-off between model accuracy and the amount of available data. We further propose data augmentation using pseudo-labels derived from less data-intensive forward surrogate models, which can often improve overall model performance. Finally, we present examples of distinct propeller geometries that exhibit nearly identical performance characteristics, illustrating the versatility and potential of GenAI in engineering design.

Liangyu Zhong, Joachim Sicking, Fabian Hüger et al.

Semantic segmentation networks have achieved significant success under the assumption of independent and identically distributed data. However, these networks often struggle to detect anomalies from unknown semantic classes due to the limited set of visual concepts they are typically trained on. To address this issue, anomaly segmentation often involves fine-tuning on outlier samples, necessitating additional efforts for data collection, labeling, and model retraining. Seeking to avoid this cumbersome work, we take a different approach and propose to incorporate Vision-Language (VL) encoders into existing anomaly detectors to leverage the semantically broad VL pre-training for improved outlier awareness. Additionally, we propose a new scoring function that enables data- and training-free outlier supervision via textual prompts. The resulting VL4AD model, which includes max-logit prompt ensembling and a class-merging strategy, achieves competitive performance on widely used benchmark datasets, thereby demonstrating the potential of vision-language models for pixel-wise anomaly detection.

Laura Fieback, Nishilkumar Balar, Jakob Spiegelberg et al.

Despite recent advances in Large Vision Language Models (LVLMs), these models still suffer from generating hallucinatory responses that do not align with the visual input provided. To mitigate such hallucinations, we introduce Efficient Contrastive Decoding (ECD), a simple method that leverages probabilistic hallucination detection to shift the output distribution towards contextually accurate answers at inference time. By contrasting token probabilities and hallucination scores, ECD subtracts hallucinated concepts from the original distribution, effectively suppressing hallucinations. Notably, our proposed method can be applied to any open-source LVLM and does not require additional LVLM training. We evaluate our method on several benchmark datasets and across different LVLMs. Our experiments show that ECD effectively mitigates hallucinations, outperforming state-of-the-art methods with respect to performance on LVLM benchmarks and computation time.

Nihal Acharya Adde, Hanno Gottschalk, Andreas Ebert

This paper focuses on hyperparameter optimization for autonomous driving strategies based on Reinforcement Learning. We provide a detailed description of training the RL agent in a simulation environment. Subsequently, we employ Efficient Global Optimization algorithm that uses Gaussian Process fitting for hyperparameter optimization in RL. Before this optimization phase, Gaussian process interpolation is applied to fit the surrogate model, for which the hyperparameter set is generated using Latin hypercube sampling. To accelerate the evaluation, parallelization techniques are employed. Following the hyperparameter optimization procedure, a set of hyperparameters is identified, resulting in a noteworthy enhancement in overall driving performance. There is a substantial increase of 4\% when compared to existing manually tuned parameters and the hyperparameters discovered during the initialization process using Latin hypercube sampling. After the optimization, we analyze the obtained results thoroughly and conduct a sensitivity analysis to assess the robustness and generalization capabilities of the learned autonomous driving strategies. The findings from this study contribute to the advancement of Gaussian process based Bayesian optimization to optimize the hyperparameters for autonomous driving in RL, providing valuable insights for the development of efficient and reliable autonomous driving systems.

Onur T. Doganay, Alexander Klawonn, Martin Eigel et al.

Partial differential equation (PDE) simulations are fundamental to engineering and physics but are often computationally prohibitive for real-time applications. While generative AI offers a promising avenue for surrogate modeling, standard video generation architectures lack the specific control and data compatibility required for physical simulations. This paper introduces a geometry aware world model architecture, derived from a video generation architecture (LongVideoGAN), designed to learn transient physics. We introduce two key architecture elements: (1) a twofold conditioning mechanism incorporating global physical parameters and local geometric masks, and (2) an architectural adaptation to support arbitrary channel dimensions, moving beyond standard RGB constraints. We evaluate this approach on a 2D transient computational fluid dynamics (CFD) problem involving convective heat transfer from buoyancy-driven flow coupled to a heat flow in a solid structure. We demonstrate that the conditioned model successfully reproduces complex temporal dynamics and spatial correlations of the training data. Furthermore, we assess the model's generalization capabilities on unseen geometric configurations, highlighting both its potential for controlled simulation synthesis and current limitations in spatial precision for out-of-distribution samples.

Patrick Krüger, Hanno Gottschalk, Werner Krebs et al.

The need to burn 100% H2 in high efficient gas turbines featuring low NOx combustion in premix mode require the complete redesign of the combustion system to ensure stable operation without any flashback. Since all engine frames featuring a power range from 4 MW up to 600 MW are affected, a huge design effort is expected. To reduce this effort, especially to transfer knowledge between the different engine classes, generative design methods using latest AI technology will provide promising potential. In this work, this challenge is approached utilizing the current advances in generative artificial intelligence. We train an Invertible Neural Network (INN) on an expandable database of geometrically parameterized combustor designs with simulated performance labels. Utilizing the INN in its inverse direction, multiple design proposals are generated which fulfill specified performance labels.

Christoph Hümmer, Manuel Schwonberg, Liangwei Zhou et al.

Domain generalization (DG) remains a significant challenge for perception based on deep neural networks (DNNs), where domain shifts occur due to synthetic data, lighting, weather, or location changes. Vision-language models (VLMs) marked a large step for the generalization capabilities and have been already applied to various tasks. Very recently, first approaches utilized VLMs for domain generalized segmentation and object detection and obtained strong generalization. However, all these approaches rely on complex modules, feature augmentation frameworks or additional models. Surprisingly and in contrast to that, we found that simple fine-tuning of vision-language pre-trained models yields competitive or even stronger generalization results while being extremely simple to apply. Moreover, we found that vision-language pre-training consistently provides better generalization than the previous standard of vision-only pre-training. This challenges the standard of using ImageNet-based transfer learning for domain generalization. Fully fine-tuning a vision-language pre-trained model is capable of reaching the domain generalization SOTA when training on the synthetic GTA5 dataset. Moreover, we confirm this observation for object detection on a novel synthetic-to-real benchmark. We further obtain superior generalization capabilities by reaching 77.9% mIoU on the popular Cityscapes-to-ACDC benchmark. We also found improved in-domain generalization, leading to an improved SOTA of 86.4% mIoU on the Cityscapes test set marking the first place on the leaderboard.

Daniel Bogdoll, Iramm Hamdard, Lukas Namgyu Rößler et al.

The scale-up of autonomous vehicles depends heavily on their ability to deal with anomalies, such as rare objects on the road. In order to handle such situations, it is necessary to detect anomalies in the first place. Anomaly detection for autonomous driving has made great progress in the past years but suffers from poorly designed benchmarks with a strong focus on camera data. In this work, we propose AnoVox, the largest benchmark for ANOmaly detection in autonomous driving to date. AnoVox incorporates large-scale multimodal sensor data and spatial VOXel ground truth, allowing for the comparison of methods independent of their used sensor. We propose a formal definition of normality and provide a compliant training dataset. AnoVox is the first benchmark to contain both content and temporal anomalies.

Miguel de Campos, Werner Krebs, Hanno Gottschalk

Inverse design aims to find design parameters $x$ achieving target performance $y^*$. Generative approaches learn bidirectional mappings between designs and labels, enabling diverse solution sampling. However, standard conditional flow matching (CFM), when adapted to inverse problems by pairing labels with design parameters, exhibits strong sensitivity to their arbitrary ordering and scaling, leading to unstable training. We introduce Diagonal Flow Matching (Diag-CFM), which resolves this through a zero-anchoring strategy that pairs design coordinates with noise and labels with zero, making the learning problem provably invariant to coordinate permutations. This yields order-of-magnitude improvements in round-trip accuracy over CFM and invertible neural network baselines across design dimensions up to $P{=}100$. We develop two architecture-intrinsic uncertainty metrics, Zero-Deviation and Self-Consistency, that enable three practical capabilities: selecting the best candidate among multiple generations, abstaining from unreliable predictions, and detecting out-of-distribution targets; consistently outperforming ensemble and general-purpose alternatives across all tasks. We validate on airfoil, gas turbine combustor, and an analytical benchmark with scalable design dimension.

Claudia Drygala, Edmund Ross, Francesca di Mare et al.

Numerical simulations of turbulent flows present significant challenges in fluid dynamics due to their complexity and high computational cost. High resolution techniques such as Direct Numerical Simulation (DNS) and Large Eddy Simulation (LES) are generally not computationally affordable, particularly for technologically relevant problems. Recent advances in machine learning, specifically in generative probabilistic models, offer promising alternatives for turbulence modeling. This paper investigates the application of three generative models - Variational Autoencoders (VAE), Deep Convolutional Generative Adversarial Networks (DCGAN), and Denoising Diffusion Probabilistic Models (DDPM) - in simulating a 2D Kármán vortex street around a fixed cylinder. Training data was obtained by means of LES. We evaluate each model's ability to capture the statistical properties and spatial structures of the turbulent flow. Our results demonstrate that DDPM and DCGAN effectively replicate the flow distribution, highlighting their potential as efficient and accurate tools for turbulence modeling. We find a strong argument for DCGAN, as although they are more difficult to train (due to problems such as mode collapse), they gave the fastest inference and training time, require less data to train compared to VAE and DDPM, and provide the results most closely aligned with the input stream. In contrast, VAE train quickly (and can generate samples quickly) but do not produce adequate results, and DDPM, whilst effective, is significantly slower at both inference and training time.

Youssef Shoeb, Azarm Nowzad, Hanno Gottschalk

In this paper, we review the state of the art in Out-of-Distribution (OoD) segmentation, with a focus on road obstacle detection in automated driving as a real-world application. We analyse the performance of existing methods on two widely used benchmarks, SegmentMeIfYouCan Obstacle Track and LostAndFound-NoKnown, highlighting their strengths, limitations, and real-world applicability. Additionally, we discuss key challenges and outline potential research directions to advance the field. Our goal is to provide researchers and practitioners with a comprehensive perspective on the current landscape of OoD segmentation and to foster further advancements toward safer and more reliable autonomous driving systems.

Youssef Shoeb, Nazir Nayal, Azarm Nowzad et al.

Detecting road obstacles is essential for autonomous vehicles to navigate dynamic and complex traffic environments safely. Current road obstacle detection methods typically assign a score to each pixel and apply a threshold to generate final predictions. However, selecting an appropriate threshold is challenging, and the per-pixel classification approach often leads to fragmented predictions with numerous false positives. In this work, we propose a novel method that leverages segment-level features from visual foundation models and likelihood ratios to predict road obstacles directly. By focusing on segments rather than individual pixels, our approach enhances detection accuracy, reduces false positives, and offers increased robustness to scene variability. We benchmark our approach against existing methods on the RoadObstacle and LostAndFound datasets, achieving state-of-the-art performance without needing a predefined threshold.

Annika Mütze, Natalie Grabowsky, Edgar Heinert et al.

Recent research has investigated the shape and texture biases of pre-trained deep neural networks (DNNs) in image classification. Those works test how much a trained DNN relies on specific image cues like texture. The present study shifts the focus to understanding the cue influence during training, analyzing what DNNs can learn from shape, texture, and color cues in absence of the others; investigating their individual and combined influence on the learning success. We analyze these cue influences at multiple levels by decomposing datasets into cue-specific versions. Addressing semantic segmentation, we learn the given task from these reduced cue datasets, creating cue experts. Early fusion of cues is performed by constructing appropriate datasets. This is complemented by a late fusion of experts which allows us to study cue influence location-dependent on pixel level. Experiments on Cityscapes, PASCAL Context, and a synthetic CARLA dataset show that while no single cue dominates, the shape + color expert predominantly improves the prediction of small objects and border pixels. The cue performance order is consistent for the tested convolutional and transformer architecture, indicating similar cue extraction capabilities, although pre-trained transformers are said to be more biased towards shape than convolutional neural networks.

Youssef Shoeb, Azarm Nowzad, Hanno Gottschalk

Advances in machine learning methods for computer vision tasks have led to their consideration for safety-critical applications like autonomous driving. However, effectively integrating these methods into the automotive development lifecycle remains challenging. Since the performance of machine learning algorithms relies heavily on the training data provided, the data and model development lifecycle play a key role in successfully integrating these components into the product development lifecycle. Existing models frequently encounter difficulties recognizing or adapting to novel instances not present in the original training dataset. This poses a significant risk for reliable deployment in dynamic environments. To address this challenge, we propose an adaptive neural network architecture and an iterative development framework that enables users to efficiently incorporate previously unknown objects into the current perception system. Our approach builds on continuous learning, emphasizing the necessity of dynamic updates to reflect real-world deployment conditions. Specifically, we introduce a pipeline with three key components: (1) a scalable network extension strategy to integrate new classes while preserving existing performance, (2) a dynamic OoD detection component that requires no additional retraining for newly added classes, and (3) a retrieval-based data augmentation process tailored for safety-critical deployments. The integration of these components establishes a pragmatic and adaptive pipeline for the continuous evolution of perception systems in the context of autonomous driving.

Katharina Winter, Abhishek Vivekanandan, Rupert Polley et al.

Generative AI (GenAI) is rapidly advancing the field of Autonomous Driving (AD), extending beyond traditional applications in text, image, and video generation. We explore how generative models can enhance automotive tasks, such as static map creation, dynamic scenario generation, trajectory forecasting, and vehicle motion planning. By examining multiple generative approaches ranging from Variational Autoencoder (VAEs) over Generative Adversarial Networks (GANs) and Invertible Neural Networks (INNs) to Generative Transformers (GTs) and Diffusion Models (DMs), we highlight and compare their capabilities and limitations for AD-specific applications. Additionally, we discuss hybrid methods integrating conventional techniques with generative approaches, and emphasize their improved adaptability and robustness. We also identify relevant datasets and outline open research questions to guide future developments in GenAI. Finally, we discuss three core challenges: safety, interpretability, and realtime capabilities, and present recommendations for image generation, dynamic scenario generation, and planning.

Claudia Drygala, Hanno Gottschalk, Thomas Kruse et al.

Brenier proved that under certain conditions on a source and a target probability measure there exists a strictly convex function such that its gradient is a transport map from the source to the target distribution. This function is called the Brenier potential. Furthermore, detailed information on the Hölder regularity of the Brenier potential is available. In this work we develop the statistical learning theory of generative adversarial neural networks that learn the Brenier potential. As by the transformation of densities formula, the density of the generated measure depends on the second derivative of the Brenier potential, we develop the universal approximation theory of ReCU networks with cubic activation $\mathtt{ReCU}(x)=\max\{0,x\}^3$ that combines the favorable approximation properties of Hölder functions with a Lipschitz continuous density. In order to assure the convexity of such general networks, we introduce an adversarial training procedure for a potential function represented by the ReCU networks that combines the classical discriminator cross entropy loss with a penalty term that enforces (strict) convexity. We give a detailed decomposition of learning errors and show that for a suitable high penalty parameter all networks chosen in the adversarial min-max optimization problem are strictly convex. This is further exploited to prove the consistency of the learning procedure for (slowly) expanding network capacity. We also implement the described learning algorithm and apply it to a number of standard test cases from Gaussian mixture to image data as target distributions. As predicted in theory, we observe that the convexity loss becomes inactive during the training process and the potentials represented by the neural networks have learned convexity.

Manuel Schwonberg, Hanno Gottschalk

The generalization of deep neural networks to unknown domains is a major challenge despite their tremendous progress in recent years. For this reason, the dynamic area of domain generalization (DG) has emerged. In contrast to unsupervised domain adaptation, there is no access to or knowledge about the target domains, and DG methods aim to generalize across multiple different unseen target domains. Domain generalization is particularly relevant for the task semantic segmentation which is used in several areas such as biomedicine or automated driving. This survey provides a comprehensive overview of the rapidly evolving topic of domain generalized semantic segmentation. We cluster and review existing approaches and identify the paradigm shift towards foundation-model-based domain generalization. Finally, we provide an extensive performance comparison of all approaches, which highlights the significant influence of foundation models on domain generalization. This survey seeks to advance domain generalization research and inspire scientists to explore new research directions.

Emily C. Ehrhardt, Hanno Gottschalk, Tobias J. Riedlinger

NeuralODE is one example for generative machine learning based on the push forward of a simple source measure with a bijective mapping, which in the case of NeuralODE is given by the flow of a ordinary differential equation. Using Liouville's formula, the log-density of the push forward measure is easy to compute and thus NeuralODE can be trained based on the maximum Likelihood method such that the Kulback-Leibler divergence between the push forward through the flow map and the target measure generating the data becomes small. In this work, we give a detailed account on the consistency of Maximum Likelihood based empirical risk minimization for a generic class of target measures. In contrast to prior work, we do not only consider the statistical learning theory, but also give a detailed numerical analysis of the NeuralODE algorithm based on the 2nd order Runge-Kutta (RK) time integration. Using the universal approximation theory for deep ReQU networks, the stability and convergence rated for the RK scheme as well as metric entropy and concentration inequalities, we are able to prove that NeuralODE is a probably approximately correct (PAC) learning algorithm.

Edmund Ross, Claudia Drygala, Leonhard Schwarz et al.

In this work, we explore the use of compact latent representations with learned time dynamics ('World Models') to simulate physical systems. Drawing on concepts from control theory, we propose a theoretical framework that explains why projecting time slices into a low-dimensional space and then concatenating to form a history ('Tokenization') is so effective at learning physics datasets, and characterise when exactly the underlying dynamics admit a reconstruction mapping from the history of previous tokenized frames to the next. To validate these claims, we develop a sequence of models with increasing complexity, starting with least-squares regression and progressing through simple linear layers, shallow adversarial learners, and ultimately full-scale generative adversarial networks (GANs). We evaluate these models on a variety of datasets, including modified forms of the heat and wave equations, the chaotic regime 2D Kuramoto-Sivashinsky equation, and a challenging computational fluid dynamics (CFD) dataset of a 2D Kármán vortex street around a fixed cylinder, where our model is successfully able to recreate the flow.

Hanno Gottschalk, Emil Partow, Tobias J. Riedlinger

This paper provides a proof of the consistency of sparse grid quadrature for numerical integration of high dimensional distributions. In a first step, a transport map is learned that normalizes the distribution to a noise distribution on the unit cube. This step is built on the statistical learning theory of neural ordinary differential equations, which has been established recently. Secondly, the composition of the generative map with the quantity of interest is integrated numerically using the Clenshaw-Curtis sparse grid quadrature. A decomposition of the total numerical error in quadrature error and statistical error is provided. As main result it is proven in the framework of empirical risk minimization that all error terms can be controlled in the sense of PAC (probably approximately correct) learning and with high probability the numerical integral approximates the theoretical value up to an arbitrary small error in the limit where the data set size is growing and the network capacity is increased adaptively.

Liangyu Zhong, Fabio Rosenthal, Joachim Sicking et al.

While Multimodal Large Language Models (MLLMs) offer strong perception and reasoning capabilities for image-text input, Visual Question Answering (VQA) focusing on small image details still remains a challenge. Although visual cropping techniques seem promising, recent approaches have several limitations: the need for task-specific fine-tuning, low efficiency due to uninformed exhaustive search, or incompatibility with efficient attention implementations. We address these shortcomings by proposing a training-free visual cropping method, dubbed FOCUS, that leverages MLLM-internal representations to guide the search for the most relevant image region. This is accomplished in four steps: first, we identify the target object(s) in the VQA prompt; second, we compute an object relevance map using the key-value (KV) cache; third, we propose and rank relevant image regions based on the map; and finally, we perform the fine-grained VQA task using the top-ranked region. As a result of this informed search strategy, FOCUS achieves strong performance across four fine-grained VQA datasets and three types of MLLMs. It outperforms three popular visual cropping methods in both accuracy and efficiency, and matches the best-performing baseline, ZoomEye, while requiring 3 - 6.5 x less compute.

Tobias J. Riedlinger, Kira Maag, Hanno Gottschalk

Deep neural networks have set the state-of-the-art in computer vision tasks such as bounding box detection and semantic segmentation. Object detectors and segmentation models assign confidence scores to predictions, reflecting the model's uncertainty in object detection or pixel-wise classification. However, these confidence estimates are often miscalibrated, as their architectures and loss functions are tailored to task performance rather than probabilistic foundation. Even with well calibrated predictions, object detectors fail to quantify uncertainty outside detected bounding boxes, i.e., the model does not make a probability assessment of whether an area without detected objects is truly free of obstacles. This poses a safety risk in applications such as automated driving, where uncertainty in empty areas remains unexplored. In this work, we propose an object detection model grounded in spatial statistics. Bounding box data matches realizations of a marked point process, commonly used to describe the probabilistic occurrence of spatial point events identified as bounding box centers, where marks are used to describe the spatial extension of bounding boxes and classes. Our statistical framework enables a likelihood-based training and provides well-defined confidence estimates for whether a region is drivable, i.e., free of objects. We demonstrate the effectiveness of our method through calibration assessments and evaluation of performance.

Annajoyce Mariani, Kira Maag, Hanno Gottschalk

Being able to generate realistic trajectory options is at the core of increasing the degree of automation of road vehicles. While model-driven, rule-based, and classical learning-based methods are widely used to tackle these tasks at present, they can struggle to effectively capture the complex, multimodal distributions of future trajectories. In this paper we investigate whether a generative adversarial network (GAN) trained on videos of bird's-eye view (BEV) traffic scenarios can generate statistically accurate trajectories that correctly capture spatial relationships between the agents. To this end, we propose a pipeline that uses low-resolution BEV occupancy grid videos as training data for a video generative model. From the generated videos of traffic scenarios we extract abstract trajectory data using single-frame object detection and frame-to-frame object matching. We particularly choose a GAN architecture for the fast training and inference times with respect to diffusion models. We obtain our best results within 100 GPU hours of training, with inference times under 20\,ms. We demonstrate the physical realism of the proposed trajectories in terms of distribution alignment of spatial and dynamic parameters with respect to the ground truth videos from the Waymo Open Motion Dataset.

Nihal Acharya Adde, Alexandra Gianzina, Hanno Gottschalk et al.

This paper introduces Evolutionary Multi-Objective Network Architecture Search (EMNAS) for the first time to optimize neural network architectures in large-scale Reinforcement Learning (RL) for Autonomous Driving (AD). EMNAS uses genetic algorithms to automate network design, tailored to enhance rewards and reduce model size without compromising performance. Additionally, parallelization techniques are employed to accelerate the search, and teacher-student methodologies are implemented to ensure scalable optimization. This research underscores the potential of transfer learning as a robust framework for optimizing performance across iterative learning processes by effectively leveraging knowledge from earlier generations to enhance learning efficiency and stability in subsequent generations. Experimental results demonstrate that tailored EMNAS outperforms manually designed models, achieving higher rewards with fewer parameters. The findings of these strategies contribute positively to EMNAS for RL in autonomous driving, advancing the field toward better-performing networks suitable for real-world scenarios.

Manuel Schwonberg, Claus Werner, Hanno Gottschalk et al.

Despite the recent progress in deep learning based computer vision, domain shifts are still one of the major challenges. Semantic segmentation for autonomous driving faces a wide range of domain shifts, e.g. caused by changing weather conditions, new geolocations and the frequent use of synthetic data in model training. Unsupervised domain adaptation (UDA) methods have emerged which adapt a model to a new target domain by only using unlabeled data of that domain. The variety of UDA methods is large but all of them use ImageNet pre-trained models. Recently, vision-language models have demonstrated strong generalization capabilities which may facilitate domain adaptation. We show that simply replacing the encoder of existing UDA methods like DACS by a vision-language pre-trained encoder can result in significant performance improvements of up to 10.0% mIoU on the GTA5-to-Cityscapes domain shift. For the generalization performance to unseen domains, the newly employed vision-language pre-trained encoder provides a gain of up to 13.7% mIoU across three unseen datasets. However, we find that not all UDA methods can be easily paired with the new encoder and that the UDA performance does not always likewise transfer into generalization performance. Finally, we perform our experiments on an adverse weather condition domain shift to further verify our findings on a pure real-to-real domain shift.

Kamil Kowol, Stefan Bracke, Hanno Gottschalk

In this study, we propose a novel approach to enrich the training data for automated driving by using a self-designed driving simulator and two human drivers to generate safety-critical corner cases in a short period of time, as already presented in~\cite{kowol22simulator}. Our results show that incorporating these corner cases during training improves the recognition of corner cases during testing, even though, they were recorded due to visual impairment. Using the corner case triggering pipeline developed in the previous work, we investigate the effectiveness of using expert models to overcome the domain gap due to different weather conditions and times of day, compared to a universal model from a development perspective. Our study reveals that expert models can provide significant benefits in terms of performance and efficiency, and can reduce the time and effort required for model training. Our results contribute to the progress of automated driving, providing a pathway for safer and more reliable autonomous vehicles on the road in the future.

Julian Burghoff, Leonhard Ackermann, Younes Salahdine et al.

In order to improve the detection and classification of malignant melanoma, this paper describes an image-based method that can achieve AUROC values of up to 0.78 without additional clinical information. Furthermore, the importance of the domain gap between two different image sources is considered, as it is important to create usability independent of hardware components such as the high-resolution scanner used. Since for the application of machine learning methods, alterations of scanner-specific properties such as brightness, contrast or sharpness can have strong (negative) effects on the quality of the prediction methods, two ways to overcome this domain gap are discussed in this paper.

Kamil Kowol, Stefan Bracke, Hanno Gottschalk

The overall goal of this work is to enrich training data for automated driving with so called corner cases. In road traffic, corner cases are critical, rare and unusual situations that challenge the perception by AI algorithms. For this purpose, we present the design of a test rig to generate synthetic corner cases using a human-in-the-loop approach. For the test rig, a real-time semantic segmentation network is trained and integrated into the driving simulation software CARLA in such a way that a human can drive on the network's prediction. In addition, a second person gets to see the same scene from the original CARLA output and is supposed to intervene with the help of a second control unit as soon as the semantic driver shows dangerous driving behavior. Interventions potentially indicate poor recognition of a critical scene by the segmentation network and then represents a corner case. In our experiments, we show that targeted enrichment of training data with corner cases leads to improvements in pedestrian detection in safety relevant episodes in road traffic.

Robin Chan, Svenja Uhlemeyer, Matthias Rottmann et al.

Semantic segmentation is a crucial component for perception in automated driving. Deep neural networks (DNNs) are commonly used for this task and they are usually trained on a closed set of object classes appearing in a closed operational domain. However, this is in contrast to the open world assumption in automated driving that DNNs are deployed to. Therefore, DNNs necessarily face data that they have never encountered previously, also known as anomalies, which are extremely safety-critical to properly cope with. In this work, we first give an overview about anomalies from an information-theoretic perspective. Next, we review research in detecting semantically unknown objects in semantic segmentation. We demonstrate that training for high entropy responses on anomalous objects outperforms other recent methods, which is in line with our theoretical findings. Moreover, we examine a method to assess the occurrence frequency of anomalies in order to select anomaly types to include into a model's set of semantic categories. We demonstrate that these anomalies can then be learned in an unsupervised fashion, which is particularly suitable in online applications based on deep learning.