J. Marius Zöllner

Martin Mayr, Marcel Dreier, Florian Kordon et al.

The imitation of cursive handwriting is mainly limited to generating handwritten words or lines. Multiple synthetic outputs must be stitched together to create paragraphs or whole pages, whereby consistency and layout information are lost. To close this gap, we propose a method for imitating handwriting at the paragraph level that also works for unseen writing styles. Therefore, we introduce a modified latent diffusion model that enriches the encoder-decoder mechanism with specialized loss functions that explicitly preserve the style and content. We enhance the attention mechanism of the diffusion model with adaptive 2D positional encoding and the conditioning mechanism to work with two modalities simultaneously: a style image and the target text. This significantly improves the realism of the generated handwriting. Our approach sets a new benchmark in our comprehensive evaluation. It outperforms all existing imitation methods at both line and paragraph levels, considering combined style and content preservation.

Svetlana Pavlitska, Nico Lambing, J. Marius Zöllner

Traffic sign recognition is an essential component of perception in autonomous vehicles, which is currently performed almost exclusively with deep neural networks (DNNs). However, DNNs are known to be vulnerable to adversarial attacks. Several previous works have demonstrated the feasibility of adversarial attacks on traffic sign recognition models. Traffic signs are particularly promising for adversarial attack research due to the ease of performing real-world attacks using printed signs or stickers. In this work, we survey existing works performing either digital or real-world attacks on traffic sign detection and classification models. We provide an overview of the latest advancements and highlight the existing research areas that require further investigation.

Daniel Grimm, Maximilian Zipfl, Felix Hertlein et al.

Precisely predicting the future trajectories of surrounding traffic participants is a crucial but challenging problem in autonomous driving, due to complex interactions between traffic agents, map context and traffic rules. Vector-based approaches have recently shown to achieve among the best performances on trajectory prediction benchmarks. These methods model simple interactions between traffic agents but don't distinguish between relation-type and attributes like their distance along the road. Furthermore, they represent lanes only by sequences of vectors representing center lines and ignore context information like lane dividers and other road elements. We present a novel approach for vector-based trajectory prediction that addresses these shortcomings by leveraging three crucial sources of information: First, we model interactions between traffic agents by a semantic scene graph, that accounts for the nature and important features of their relation. Second, we extract agent-centric image-based map features to model the local map context. Finally, we generate anchor paths to enforce the policy in multi-modal prediction to permitted trajectories only. Each of these three enhancements shows advantages over the baseline model HoliGraph.

Svetlana Pavlitskaya, Bianca-Marina Codău, J. Marius Zöllner

Standard approaches for adversarial patch generation lead to noisy conspicuous patterns, which are easily recognizable by humans. Recent research has proposed several approaches to generate naturalistic patches using generative adversarial networks (GANs), yet only a few of them were evaluated on the object detection use case. Moreover, the state of the art mostly focuses on suppressing a single large bounding box in input by overlapping it with the patch directly. Suppressing objects near the patch is a different, more complex task. In this work, we have evaluated the existing approaches to generate inconspicuous patches. We have adapted methods, originally developed for different computer vision tasks, to the object detection use case with YOLOv3 and the COCO dataset. We have evaluated two approaches to generate naturalistic patches: by incorporating patch generation into the GAN training process and by using the pretrained GAN. For both cases, we have assessed a trade-off between performance and naturalistic patch appearance. Our experiments have shown, that using a pre-trained GAN helps to gain realistic-looking patches while preserving the performance similar to conventional adversarial patches.

Daniel Bogdoll, Svenja Uhlemeyer, Kamil Kowol et al.

Deep neural networks (DNN) which are employed in perception systems for autonomous driving require a huge amount of data to train on, as they must reliably achieve high performance in all kinds of situations. However, these DNN are usually restricted to a closed set of semantic classes available in their training data, and are therefore unreliable when confronted with previously unseen instances. Thus, multiple perception datasets have been created for the evaluation of anomaly detection methods, which can be categorized into three groups: real anomalies in real-world, synthetic anomalies augmented into real-world and completely synthetic scenes. This survey provides a structured and, to the best of our knowledge, complete overview and comparison of perception datasets for anomaly detection in autonomous driving. Each chapter provides information about tasks and ground truth, context information, and licenses. Additionally, we discuss current weaknesses and gaps in existing datasets to underline the importance of developing further data.

Larissa T. Triess, Christoph B. Rist, David Peter et al.

A considerable amount of research is concerned with the generation of realistic sensor data. LiDAR point clouds are generated by complex simulations or learned generative models. The generated data is usually exploited to enable or improve downstream perception algorithms. Two major questions arise from these procedures: First, how to evaluate the realism of the generated data? Second, does more realistic data also lead to better perception performance? This paper addresses both questions and presents a novel metric to quantify the realism of LiDAR point clouds. Relevant features are learned from real-world and synthetic point clouds by training on a proxy classification task. In a series of experiments, we demonstrate the application of our metric to determine the realism of generated LiDAR data and compare the realism estimation of our metric to the performance of a segmentation model. We confirm that our metric provides an indication for the downstream segmentation performance.

Marc Uecker, Tobias Fleck, Marcel Pflugfelder et al.

LiDAR sensors are an integral part of modern autonomous vehicles as they provide an accurate, high-resolution 3D representation of the vehicle's surroundings. However, it is computationally difficult to make use of the ever-increasing amounts of data from multiple high-resolution LiDAR sensors. As frame-rates, point cloud sizes and sensor resolutions increase, real-time processing of these point clouds must still extract semantics from this increasingly precise picture of the vehicle's environment. One deciding factor of the run-time performance and accuracy of deep neural networks operating on these point clouds is the underlying data representation and the way it is computed. In this work, we examine the relationship between the computational representations used in neural networks and their performance characteristics. To this end, we propose a novel computational taxonomy of LiDAR point cloud representations used in modern deep neural networks for 3D point cloud processing. Using this taxonomy, we perform a structured analysis of different families of approaches. Thereby, we uncover common advantages and limitations in terms of computational efficiency, memory requirements, and representational capacity as measured by semantic segmentation performance. Finally, we provide some insights and guidance for future developments in neural point cloud processing methods.

Faris Janjoš, Max Keller, Maxim Dolgov et al.

Accurate vehicle trajectory prediction is an unsolved problem in autonomous driving with various open research questions. State-of-the-art approaches regress trajectories either in a one-shot or step-wise manner. Although one-shot approaches are usually preferred for their simplicity, they relinquish powerful self-supervision schemes that can be constructed by chaining multiple time-steps. We address this issue by proposing a middle-ground where multiple trajectory segments are chained together. Our proposed Multi-Branch Self-Supervised Predictor receives additional training on new predictions starting at intermediate future segments. In addition, the model 'imagines' the latent context and 'predicts the past' while combining multi-modal trajectories in a tree-like manner. We deliberately keep aspects such as interaction and environment modeling simplistic and nevertheless achieve competitive results on the INTERACTION dataset. Furthermore, we investigate the sparsely explored uncertainty estimation of deterministic predictors. We find positive correlations between the prediction error and two proposed metrics, which might pave way for determining prediction confidence.

Svetlana Pavlitska, Nico Lambing, Ashok Kumar Bangaru et al.

Real-time traffic light recognition is essential for autonomous driving. Yet, a cohesive overview of the underlying model architectures for this task is currently missing. In this work, we conduct a comprehensive survey and analysis of traffic light recognition methods that use convolutional neural networks (CNNs). We focus on two essential aspects: datasets and CNN architectures. Based on an underlying architecture, we cluster methods into three major groups: (1) modifications of generic object detectors which compensate for specific task characteristics, (2) multi-stage approaches involving both rule-based and CNN components, and (3) task-specific single-stage methods. We describe the most important works in each cluster, discuss the usage of the datasets, and identify research gaps.

Svetlana Pavlitskaya, Joël Oswald, J. Marius Zöllner

Although numerous methods to reduce the overfitting of convolutional neural networks (CNNs) exist, it is still not clear how to confidently measure the degree of overfitting. A metric reflecting the overfitting level might be, however, extremely helpful for the comparison of different architectures and for the evaluation of various techniques to tackle overfitting. Motivated by the fact that overfitted neural networks tend to rather memorize noise in the training data than generalize to unseen data, we examine how the training accuracy changes in the presence of increasing data perturbations and study the connection to overfitting. While previous work focused on label noise only, we examine a spectrum of techniques to inject noise into the training data, including adversarial perturbations and input corruptions. Based on this, we define two new metrics that can confidently distinguish between correct and overfitted models. For the evaluation, we derive a pool of models for which the overfitting behavior is known beforehand. To test the effect of various factors, we introduce several anti-overfitting measures in architectures based on VGG and ResNet and study their impact, including regularization techniques, training set size, and the number of parameters. Finally, we assess the applicability of the proposed metrics by measuring the overfitting degree of several CNN architectures outside of our model pool.

Svetlana Pavlitskaya, Nikolai Polley, Michael Weber et al.

Taking into account information across the temporal domain helps to improve environment perception in autonomous driving. However, it has not been studied so far whether temporally fused neural networks are vulnerable to deliberately generated perturbations, i.e. adversarial attacks, or whether temporal history is an inherent defense against them. In this work, we study whether temporal feature networks for object detection are vulnerable to universal adversarial attacks. We evaluate attacks of two types: imperceptible noise for the whole image and locally-bound adversarial patch. In both cases, perturbations are generated in a white-box manner using PGD. Our experiments confirm, that attacking even a portion of a temporal input suffices to fool the network. We visually assess generated perturbations to gain insights into the functioning of attacks. To enhance the robustness, we apply adversarial training using 5-PGD. Our experiments on KITTI and nuScenes datasets demonstrate, that a model robustified via K-PGD is able to withstand the studied attacks while keeping the mAP-based performance comparable to that of an unattacked model.

Svetlana Pavlitska, Hannes Grolig, J. Marius Zöllner

Increasing the model capacity is a known approach to enhance the adversarial robustness of deep learning networks. On the other hand, various model compression techniques, including pruning and quantization, can reduce the size of the network while preserving its accuracy. Several recent studies have addressed the relationship between model compression and adversarial robustness, while some experiments have reported contradictory results. This work summarizes available evidence and discusses possible explanations for the observed effects.

Abhishek Vivekanandan, Ahmed Abouelazm, Philip Schörner et al.

Accurately forecasting the motion of traffic actors is crucial for the deployment of autonomous vehicles at a large scale. Current trajectory forecasting approaches primarily concentrate on optimizing a loss function with a specific metric, which can result in predictions that do not adhere to physical laws or violate external constraints. Our objective is to incorporate explicit knowledge priors that allow a network to forecast future trajectories in compliance with both the kinematic constraints of a vehicle and the geometry of the driving environment. To achieve this, we introduce a non-parametric pruning layer and attention layers to integrate the defined knowledge priors. Our proposed method is designed to ensure reachability guarantees for traffic actors in both complex and dynamic situations. By conditioning the network to follow physical laws, we can obtain accurate and safe predictions, essential for maintaining autonomous vehicles' safety and efficiency in real-world settings.In summary, this paper presents concepts that prevent off-road predictions for safe and reliable motion forecasting by incorporating knowledge priors into the training process.

Svetlana Pavlitskaya, Jonas Hendl, Sebastian Kleim et al.

Adversarial patch-based attacks aim to fool a neural network with an intentionally generated noise, which is concentrated in a particular region of an input image. In this work, we perform an in-depth analysis of different patch generation parameters, including initialization, patch size, and especially positioning a patch in an image during training. We focus on the object vanishing attack and run experiments with YOLOv3 as a model under attack in a white-box setting and use images from the COCO dataset. Our experiments have shown, that inserting a patch inside a window of increasing size during training leads to a significant increase in attack strength compared to a fixed position. The best results were obtained when a patch was positioned randomly during training, while patch position additionally varied within a batch.

Daniel Bogdoll, Lukas Bosch, Tim Joseph et al.

In recent years there have been remarkable advancements in autonomous driving. While autonomous vehicles demonstrate high performance in closed-set conditions, they encounter difficulties when confronted with unexpected situations. At the same time, world models emerged in the field of model-based reinforcement learning as a way to enable agents to predict the future depending on potential actions. This led to outstanding results in sparse reward and complex control tasks. This work provides an overview of how world models can be leveraged to perform anomaly detection in the domain of autonomous driving. We provide a characterization of world models and relate individual components to previous works in anomaly detection to facilitate further research in the field.

Philipp Stegmaier, Karl Kurzer, J. Marius Zöllner

Automated vehicles require the ability to cooperate with humans for smooth integration into today's traffic. While the concept of cooperation is well known, developing a robust and efficient cooperative trajectory planning method is still a challenge. One aspect of this challenge is the uncertainty surrounding the state of the environment due to limited sensor accuracy. This uncertainty can be represented by a Partially Observable Markov Decision Process. Our work addresses this problem by extending an existing cooperative trajectory planning approach based on Monte Carlo Tree Search for continuous action spaces. It does so by explicitly modeling uncertainties in the form of a root belief state, from which start states for trees are sampled. After the trees have been constructed with Monte Carlo Tree Search, their results are aggregated into return distributions using kernel regression. We apply two risk metrics for the final selection, namely a Lower Confidence Bound and a Conditional Value at Risk. It can be demonstrated that the integration of risk metrics in the final selection policy consistently outperforms a baseline in uncertain environments, generating considerably safer trajectories.

Niklas Wagner, Felix Mätzler, Samed R. Vossberg et al.

Understanding emotions and expressions is a task of interest across multiple disciplines, especially for improving user experiences. Contrary to the common perception, it has been shown that emotions are not discrete entities but instead exist along a continuum. People understand discrete emotions differently due to a variety of factors, including cultural background, individual experiences, and cognitive biases. Therefore, most approaches to expression understanding, particularly those relying on discrete categories, are inherently biased. In this paper, we present a comparative in-depth analysis of two common datasets (AffectNet and EMOTIC) equipped with the components of the circumplex model of affect. Further, we propose a model for the prediction of facial expressions tailored for lightweight applications. Using a small-scaled MaxViT-based model architecture, we evaluate the impact of discrete expression category labels in training with the continuous valence and arousal labels. We show that considering valence and arousal in addition to discrete category labels helps to significantly improve expression inference. The proposed model outperforms the current state-of-the-art models on AffectNet, establishing it as the best-performing model for inferring valence and arousal achieving a 7% lower RMSE. Training scripts and trained weights to reproduce our results can be found here: https://github.com/wagner-niklas/CAGE_expression_inference.

Svetlana Pavlitska, Oliver Bagge, Federico Peccia et al.

Model compression and hardware acceleration are essential for the resource-efficient deployment of deep neural networks. Modern object detectors have highly interconnected convolutional layers with concatenations. In this work, we study how pruning can be applied to such architectures, exemplary for YOLOv7. We propose a method to handle concatenation layers, based on the connectivity graph of convolutional layers. By automating iterative sensitivity analysis, pruning, and subsequent model fine-tuning, we can significantly reduce model size both in terms of the number of parameters and FLOPs, while keeping comparable model accuracy. Finally, we deploy pruned models to FPGA and NVIDIA Jetson Xavier AGX. Pruned models demonstrate a 2x speedup for the convolutional layers in comparison to the unpruned counterparts and reach real-time capability with 14 FPS on FPGA. Our code is available at https://github.com/fzi-forschungszentrum-informatik/iterative-yolo-pruning.

Marc Uecker, J. Marius Zöllner

Recently, LiDAR segmentation methods for autonomous vehicles, powered by deep neural networks, have experienced steep growth in performance on classic benchmarks, such as nuScenes and SemanticKITTI. However, there are still large gaps in performance when deploying models trained on such single-sensor setups to modern vehicles with multiple high-resolution LiDAR sensors. In this work, we introduce a new metric for feature-level invariance which can serve as a proxy to measure cross-domain generalization without requiring labeled data. Additionally, we propose two application-specific data augmentations, which facilitate better transfer to multi-sensor LiDAR setups, when trained on single-sensor datasets. We provide experimental evidence on both simulated and real data, that our proposed augmentations improve invariance across LiDAR setups, leading to improved generalization.

Svetlana Pavlitska, Beyza Keskin, Alwin Faßbender et al.

Estimating accurate and well-calibrated predictive uncertainty is important for enhancing the reliability of computer vision models, especially in safety-critical applications like traffic scene perception. While ensemble methods are commonly used to quantify uncertainty by combining multiple models, a mixture of experts (MoE) offers an efficient alternative by leveraging a gating network to dynamically weight expert predictions based on the input. Building on the promising use of MoEs for semantic segmentation in our previous works, we show that well-calibrated predictive uncertainty estimates can be extracted from MoEs without architectural modifications. We investigate three methods to extract predictive uncertainty estimates: predictive entropy, mutual information, and expert variance. We evaluate these methods for an MoE with two experts trained on a semantical split of the A2D2 dataset. Our results show that MoEs yield more reliable uncertainty estimates than ensembles in terms of conditional correctness metrics under out-of-distribution (OOD) data. Additionally, we evaluate routing uncertainty computed via gate entropy and find that simple gating mechanisms lead to better calibration of routing uncertainty estimates than more complex classwise gates. Finally, our experiments on the Cityscapes dataset suggest that increasing the number of experts can further enhance uncertainty calibration. Our code is available at https://github.com/KASTEL-MobilityLab/mixtures-of-experts/.

Svetlana Pavlitska, Enrico Eisen, J. Marius Zöllner

Vulnerability to adversarial attacks is a well-known deficiency of deep neural networks. Larger networks are generally more robust, and ensembling is one method to increase adversarial robustness: each model's weaknesses are compensated by the strengths of others. While an ensemble uses a deterministic rule to combine model outputs, a mixture of experts (MoE) includes an additional learnable gating component that predicts weights for the outputs of the expert models, thus determining their contributions to the final prediction. MoEs have been shown to outperform ensembles on specific tasks, yet their susceptibility to adversarial attacks has not been studied yet. In this work, we evaluate the adversarial vulnerability of MoEs for semantic segmentation of urban and highway traffic scenes. We show that MoEs are, in most cases, more robust to per-instance and universal white-box adversarial attacks and can better withstand transfer attacks. Our code is available at \url{https://github.com/KASTEL-MobilityLab/mixtures-of-experts/}.

Svetlana Pavlitska, Leopold Müller, J. Marius Zöllner

Adversarial attacks on traffic sign classification models were among the first successfully tried in the real world. Since then, the research in this area has been mainly restricted to repeating baseline models, such as LISA-CNN or GTSRB-CNN, and similar experiment settings, including white and black patches on traffic signs. In this work, we decouple model architectures from the datasets and evaluate on further generic models to make a fair comparison. Furthermore, we compare two attack settings, inconspicuous and visible, which are usually regarded without direct comparison. Our results show that standard baselines like LISA-CNN or GTSRB-CNN are significantly more susceptible than the generic ones. We, therefore, suggest evaluating new attacks on a broader spectrum of baselines in the future. Our code is available at \url{https://github.com/KASTEL-MobilityLab/attacks-on-traffic-sign-recognition/}.

Abhishek Vivekanandan, J. Marius Zöllner

Representing diverse and plausible future trajectories is critical for motion forecasting in autonomous driving. However, efficiently capturing these trajectories in a compact set remains challenging. This study introduces a novel approach for generating scene-specific trajectory sets tailored to different contexts, such as intersections and straight roads, by leveraging map information and actor dynamics. A deterministic goal sampling algorithm identifies relevant map regions, while our Recursive In-Distribution Subsampling (RIDS) method enhances trajectory plausibility by condensing redundant representations. Experiments on the Argoverse 2 dataset demonstrate that our method achieves up to a 10% improvement in Driving Area Compliance (DAC) compared to baseline methods while maintaining competitive displacement errors. Our work highlights the benefits of mining such scene-aware trajectory sets and how they could capture the complex and heterogeneous nature of actor behavior in real-world driving scenarios.

Svetlana Pavlitska, Haixi Fan, Konstantin Ditschuneit et al.

Robustifying convolutional neural networks (CNNs) against adversarial attacks remains challenging and often requires resource-intensive countermeasures. We explore the use of sparse mixture-of-experts (MoE) layers to improve robustness by replacing selected residual blocks or convolutional layers, thereby increasing model capacity without additional inference cost. On ResNet architectures trained on CIFAR-100, we find that inserting a single MoE layer in the deeper stages leads to consistent improvements in robustness under PGD and AutoPGD attacks when combined with adversarial training. Furthermore, we discover that when switch loss is used for balancing, it causes routing to collapse onto a small set of overused experts, thereby concentrating adversarial training on these paths and inadvertently making them more robust. As a result, some individual experts outperform the gated MoE model in robustness, suggesting that robust subpaths emerge through specialization. Our code is available at https://github.com/KASTEL-MobilityLab/robust-sparse-moes.

Daniel Bogdoll, Jing Qin, Moritz Nekolla et al.

Reinforcement Learning is a highly active research field with promising advancements. In the field of autonomous driving, however, often very simple scenarios are being examined. Common approaches use non-interpretable control commands as the action space and unstructured reward designs which lack structure. In this work, we introduce Informed Reinforcement Learning, where a structured rulebook is integrated as a knowledge source. We learn trajectories and asses them with a situation-aware reward design, leading to a dynamic reward which allows the agent to learn situations which require controlled traffic rule exceptions. Our method is applicable to arbitrary RL models. We successfully demonstrate high completion rates of complex scenarios with recent model-based agents.

Rupert Polley, Nikolai Polley, Dominik Heid et al.

Traffic light perception is an essential component of the camera-based perception system for autonomous vehicles, enabling accurate detection and interpretation of traffic lights to ensure safe navigation through complex urban environments. In this work, we propose a modularized perception framework that integrates state-of-the-art detection models with a novel real-time association and decision framework, enabling seamless deployment into an autonomous driving stack. To address the limitations of existing public datasets, we introduce the ATLAS dataset, which provides comprehensive annotations of traffic light states and pictograms across diverse environmental conditions and camera setups. This dataset is publicly available at https://url.fzi.de/ATLAS. We train and evaluate several state-of-the-art traffic light detection architectures on ATLAS, demonstrating significant performance improvements in both accuracy and robustness. Finally, we evaluate the framework in real-world scenarios by deploying it in an autonomous vehicle to make decisions at traffic light-controlled intersections, highlighting its reliability and effectiveness for real-time operation.

Marc Uecker, J. Marius Zöllner

In this paper, we introduce a method for estimating blind spots for sensor setups of autonomous or automated vehicles and/or robotics applications. In comparison to previous methods that rely on geometric approximations, our presented approach provides more realistic coverage estimates by utilizing accurate and detailed 3D simulation environments. Our method leverages point clouds from LiDAR sensors or camera depth images from high-fidelity simulations of target scenarios to provide accurate and actionable visibility estimates. A Monte Carlo-based reference sensor simulation enables us to accurately estimate blind spot size as a metric of coverage, as well as detection probabilities of objects at arbitrary positions.

Melih Yazgan, Allen Xavier Arasan, J. Marius Zöllner

Collaborative perception allows connected vehicles to exchange sensor information and overcome each vehicle's blind spots. Yet transmitting raw point clouds or full feature maps overwhelms Vehicle-to-Vehicle (V2V) communications, causing latency and scalability problems. We introduce EffiComm, an end-to-end framework that transmits less than 40% of the data required by prior art while maintaining state-of-the-art 3D object detection accuracy. EffiComm operates on Bird's-Eye-View (BEV) feature maps from any modality and applies a two-stage reduction pipeline: (1) Selective Transmission (ST) prunes low-utility regions with a confidence mask; (2) Adaptive Grid Reduction (AGR) uses a Graph Neural Network (GNN) to assign vehicle-specific keep ratios according to role and network load. The remaining features are fused with a soft-gated Mixture-of-Experts (MoE) attention layer, offering greater capacity and specialization for effective feature integration. On the OPV2V benchmark, EffiComm reaches 0.84 mAP@0.7 while sending only an average of approximately 1.5 MB per frame, outperforming previous methods on the accuracy-per-bit curve. These results highlight the value of adaptive, learned communication for scalable Vehicle-to-Everything (V2X) perception.

Nikolai Polley, Yacin Boualili, Ferdinand Mütsch et al.

On-board sensors of autonomous vehicles can be obstructed, occluded, or limited by restricted fields of view, complicating downstream driving decisions. Intelligent roadside infrastructure perception systems, installed at elevated vantage points, can provide wide, unobstructed intersection coverage, supplying a complementary information stream to autonomous vehicles via vehicle-to-everything (V2X) communication. However, conventional 3D object-detection algorithms struggle to generalize under the domain shift introduced by top-down perspectives and steep camera angles. We introduce a 2.5D object detection framework, tailored specifically for infrastructure roadside-mounted cameras. Unlike conventional 2D or 3D object detection, we employ a prediction approach to detect ground planes of vehicles as parallelograms in the image frame. The parallelogram preserves the planar position, size, and orientation of objects while omitting their height, which is unnecessary for most downstream applications. For training, a mix of real-world and synthetically generated scenes is leveraged. We evaluate generalizability on a held-out camera viewpoint and in adverse-weather scenarios absent from the training set. Our results show high detection accuracy, strong cross-viewpoint generalization, and robustness to diverse lighting and weather conditions. Model weights and inference code are provided at: https://gitlab.kit.edu/kit/aifb/ATKS/public/digit4taf/2.5d-object-detection

Daniel Grimm, Ahmed Abouelazm, J. Marius Zöllner

To achieve full autonomous driving, a good understanding of the surrounding environment is necessary. Especially predicting the future states of other traffic participants imposes a non-trivial challenge. Current SotA-models already show promising results when trained on real datasets (e.g. Argoverse2, NuScenes). Problems arise when these models are deployed to new/unseen areas. Typically, performance drops significantly, indicating that the models lack generalization. In this work, we introduce a new Graph Neural Network (GNN) that utilizes a heterogeneous graph consisting of traffic participants and vectorized road network. Latter, is used to classify goals, i.e. endpoints of the predicted trajectories, in a multi-staged approach, leading to a better generalization to unseen scenarios. We show the effectiveness of the goal selection process via cross-dataset evaluation, i.e. training on Argoverse2 and evaluating on NuScenes.

Abhishek Vivekanandan, Christian Hubschneider, J. Marius Zöllner

The ability to retrieve semantically and directionally similar short-range trajectories with both accuracy and efficiency is foundational for downstream applications such as motion forecasting and autonomous navigation. However, prevailing approaches often depend on computationally intensive heuristics or latent anchor representations that lack interpretability and controllability. In this work, we propose a novel framework for learning fixed-dimensional embeddings for short trajectories by leveraging a Transformer encoder trained with a contrastive triplet loss that emphasize the importance of discriminative feature spaces for trajectory data. We analyze the influence of Cosine and FFT-based similarity metrics within the contrastive learning paradigm, with a focus on capturing the nuanced directional intent that characterizes short-term maneuvers. Our empirical evaluation on the Argoverse 2 dataset demonstrates that embeddings shaped by Cosine similarity objectives yield superior clustering of trajectories by both semantic and directional attributes, outperforming FFT-based baselines in retrieval tasks. Notably, we show that compact Transformer architectures, even with low-dimensional embeddings (e.g., 16 dimensions, but qualitatively down to 4), achieve a compelling balance between retrieval performance (minADE, minFDE) and computational overhead, aligning with the growing demand for scalable and interpretable motion priors in real-time systems. The resulting embeddings provide a compact, semantically meaningful, and efficient representation of trajectory data, offering a robust alternative to heuristic similarity measures and paving the way for more transparent and controllable motion forecasting pipelines.

Rupert Polley, Sai Vignesh Abishek Deenadayalan, J. Marius Zöllner

Deep neural networks for aerial image segmentation require large amounts of labeled data, but high-quality aerial datasets with precise annotations are scarce and costly to produce. To address this limitation, we propose a self-supervised pretraining method that improves segmentation performance while reducing reliance on labeled data. Our approach uses inpainting-based pretraining, where the model learns to reconstruct missing regions in aerial images, capturing their inherent structure before being fine-tuned for road extraction. This method improves generalization, enhances robustness to domain shifts, and is invariant to model architecture and dataset choice. Experiments show that our pretraining significantly boosts segmentation accuracy, especially in low-data regimes, making it a scalable solution for aerial image analysis.

Ferdinand Mütsch, Helen Gremmelmaier, Nicolas Becker et al.

Simulation is an integral part in the process of developing autonomous vehicles and advantageous for training, validation, and verification of driving functions. Even though simulations come with a series of benefits compared to real-world experiments, various challenges still prevent virtual testing from entirely replacing physical test-drives. Our work provides an overview of these challenges with regard to different aspects and types of simulation and subsumes current trends to overcome them. We cover aspects around perception-, behavior- and content-realism as well as general hurdles in the domain of simulation. Among others, we observe a trend of data-driven, generative approaches and high-fidelity data synthesis to increasingly replace model-based simulation.

Karl Kurzer, Matthias Bitzer, J. Marius Zöllner

Cooperative trajectory planning methods for automated vehicles can solve traffic scenarios that require a high degree of cooperation between traffic participants. However, for cooperative systems to integrate into human-centered traffic, the automated systems must behave human-like so that humans can anticipate the system's decisions. While Reinforcement Learning has made remarkable progress in solving the decision-making part, it is non-trivial to parameterize a reward model that yields predictable actions. This work employs feature-based Maximum Entropy Inverse Reinforcement Learning combined with Monte Carlo Tree Search to learn reward models that maximize the likelihood of recorded multi-agent cooperative expert trajectories. The evaluation demonstrates that the approach can recover a reasonable reward model that mimics the expert and performs similarly to a manually tuned baseline reward model.

Daniel Bogdoll, Moritz Nekolla, Tim Joseph et al.

Driving on roads is restricted by various traffic rules, aiming to ensure safety for all traffic participants. However, human road users usually do not adhere to these rules strictly, resulting in varying degrees of rule conformity. Such deviations from given rules are key components of today's road traffic. In autonomous driving, robotic agents can disturb traffic flow, when rule deviations are not taken into account. In this paper, we present an approach to derive the distribution of degrees of rule conformity from human driving data. We demonstrate our method with the Waymo Open Motion dataset and Safety Distance and Speed Limit rules.

Larissa T. Triess, David Peter, J. Marius Zöllner

A number of applications, such as mobile robots or automated vehicles, use LiDAR sensors to obtain detailed information about their three-dimensional surroundings. Many methods use image-like projections to efficiently process these LiDAR measurements and use deep convolutional neural networks to predict semantic classes for each point in the scan. The spatial stationary assumption enables the usage of convolutions. However, LiDAR scans exhibit large differences in appearance over the vertical axis. Therefore, we propose semi local convolution (SLC), a convolution layer with reduced amount of weight-sharing along the vertical dimension. We are first to investigate the usage of such a layer independent of any other model changes. Our experiments did not show any improvement over traditional convolution layers in terms of segmentation IoU or accuracy.

Faris Janjoš, Maxim Dolgov, J. Marius Zöllner

In this work, we present a novel multi-modal multi-agent trajectory prediction architecture, focusing on map and interaction modeling using graph representation. For the purposes of map modeling, we capture rich topological structure into vector-based star graphs, which enable an agent to directly attend to relevant regions along polylines that are used to represent the map. We denote this architecture StarNet, and integrate it in a single-agent prediction setting. As the main result, we extend this architecture to joint scene-level prediction, which produces multiple agents' predictions simultaneously. The key idea in joint-StarNet is integrating the awareness of one agent in its own reference frame with how it is perceived from the points of view of other agents. We achieve this via masked self-attention. Both proposed architectures are built on top of the action-space prediction framework introduced in our previous work, which ensures kinematically feasible trajectory predictions. We evaluate the methods on the interaction-rich inD and INTERACTION datasets, with both StarNet and joint-StarNet achieving improvements over state of the art.

Maximilian Zipfl, J. Marius Zöllner

For the classification of traffic scenes, a description model is necessary that can describe the scene in a uniform way, independent of its domain. A model to describe a traffic scene in a semantic way is described in this paper. The description model allows to describe a traffic scene independently of the road geometry and road topology. Here, the traffic participants are projected onto the road network and represented as nodes in a graph. Depending on the relative location between two traffic participants with respect to the road topology, semantically classified edges are created between the corresponding nodes. For concretization, the edge attributes are extended by relative distances and velocities between both traffic participants with regard to the course of the lane. An important aspect of the description is that it can be converted easily into a machine-readable format. The current description focuses on dynamic objects of a traffic scene and considers traffic participants, such as pedestrians or vehicles.

Daniel Bogdoll, Johannes Jestram, Jonas Rauch et al.

In the foreseeable future, autonomous vehicles will require human assistance in situations they can not resolve on their own. In such scenarios, remote assistance from a human can provide the required input for the vehicle to continue its operation. Typical sensors used in autonomous vehicles include camera and lidar sensors. Due to the massive volume of sensor data that must be sent in real-time, highly efficient data compression is elementary to prevent an overload of network infrastructure. Sensor data compression using deep generative neural networks has been shown to outperform traditional compression approaches for both image and lidar data, regarding compression rate as well as reconstruction quality. However, there is a lack of research about the performance of generative-neural-network-based compression algorithms for remote assistance. In order to gain insights into the feasibility of deep generative models for usage in remote assistance, we evaluate state-of-the-art algorithms regarding their applicability and identify potential weaknesses. Further, we implement an online pipeline for processing sensor data and demonstrate its performance for remote assistance using the CARLA simulator.

Larissa T. Triess, David Peter, Stefan A. Baur et al.

Judging the quality of samples synthesized by generative models can be tedious and time consuming, especially for complex data structures, such as point clouds. This paper presents a novel approach to quantify the realism of local regions in LiDAR point clouds. Relevant features are learned from real-world and synthetic point clouds by training on a proxy classification task. Inspired by fair networks, we use an adversarial technique to discourage the encoding of dataset-specific information. The resulting metric can assign a quality score to samples without requiring any task specific annotations. In a series of experiments, we confirm the soundness of our metric by applying it in controllable task setups and on unseen data. Additional experiments show reliable interpolation capabilities of the metric between data with varying degree of realism. As one important application, we demonstrate how the local realism score can be used for anomaly detection in point clouds.

Faris Janjoš, Maxim Dolgov, J. Marius Zöllner

Making informed driving decisions requires reliable prediction of other vehicles' trajectories. In this paper, we present a novel learned multi-modal trajectory prediction architecture for automated driving. It achieves kinematically feasible predictions by casting the learning problem into the space of accelerations and steering angles -- by performing action-space prediction, we can leverage valuable model knowledge. Additionally, the dimensionality of the action manifold is lower than that of the state manifold, whose intrinsically correlated states are more difficult to capture in a learned manner. For the purpose of action-space prediction, we present the simple Feed-Forward Action-Space Prediction (FFW-ASP) architecture. Then, we build on this notion and introduce the novel Self-Supervised Action-Space Prediction (SSP-ASP) architecture that outputs future environment context features in addition to trajectories. A key element in the self-supervised architecture is that, based on an observed action history and past context features, future context features are predicted prior to future trajectories. The proposed methods are evaluated on real-world datasets containing urban intersections and roundabouts, and show accurate predictions, outperforming state-of-the-art for kinematically feasible predictions in several prediction metrics.

Daniel Bogdoll, Stefan Orf, Lars Töttel et al.

Corner cases for driving automation systems can often be detected by the system itself and subsequently resolved by remote humans. There exists a wide variety of technical approaches on how remote humans can resolve such issues. Over multiple domains, no common taxonomy on those approaches has developed yet, though. As the scaling of automated driving systems continues to increase, a uniform taxonomy is desirable to improve communication within the scientific community, but also beyond to policymakers and the general public. In this paper, we provide a survey on recent terminologies and propose a taxonomy for remote human input systems, classifying the different approaches based on their complexity.

Larissa T. Triess, Mariella Dreissig, Christoph B. Rist et al.

Scalable systems for automated driving have to reliably cope with an open-world setting. This means, the perception systems are exposed to drastic domain shifts, like changes in weather conditions, time-dependent aspects, or geographic regions. Covering all domains with annotated data is impossible because of the endless variations of domains and the time-consuming and expensive annotation process. Furthermore, fast development cycles of the system additionally introduce hardware changes, such as sensor types and vehicle setups, and the required knowledge transfer from simulation. To enable scalable automated driving, it is therefore crucial to address these domain shifts in a robust and efficient manner. Over the last years, a vast amount of different domain adaptation techniques evolved. There already exists a number of survey papers for domain adaptation on camera images, however, a survey for LiDAR perception is absent. Nevertheless, LiDAR is a vital sensor for automated driving that provides detailed 3D scans of the vehicle's surroundings. To stimulate future research, this paper presents a comprehensive review of recent progress in domain adaptation methods and formulates interesting research questions specifically targeted towards LiDAR perception.

Hannes Reichert, Lukas Lang, Kevin Rösch et al.

Full-stack autonomous driving perception modules usually consist of data-driven models based on multiple sensor modalities. However, these models might be biased to the sensor setup used for data acquisition. This bias can seriously impair the perception models' transferability to new sensor setups, which continuously occur due to the market's competitive nature. We envision sensor data abstraction as an interface between sensor data and machine learning applications for highly automated vehicles (HAD). For this purpose, we review the primary sensor modalities, camera, lidar, and radar, published in autonomous-driving related datasets, examine single sensor abstraction and abstraction of sensor setups, and identify critical paths towards an abstraction of sensor data from multiple perception configurations.

Michael Weber, Tassilo Wald, J. Marius Zöllner

For reliable environment perception, the use of temporal information is essential in some situations. Especially for object detection, sometimes a situation can only be understood in the right perspective through temporal information. Since image-based object detectors are currently based almost exclusively on CNN architectures, an extension of their feature extraction with temporal features seems promising. Within this work we investigate different architectural components for a CNN-based temporal information extraction. We present a Temporal Feature Network which is based on the insights gained from our architectural investigations. This network is trained from scratch without any ImageNet information based pre-training as these images are not available with temporal information. The object detector based on this network is evaluated against the non-temporal counterpart as baseline and achieves competitive results in an evaluation on the KITTI object detection dataset.

Karl Kurzer, Philip Schörner, Alexander Albers et al.

Data driven approaches for decision making applied to automated driving require appropriate generalization strategies, to ensure applicability to the world's variability. Current approaches either do not generalize well beyond the training data or are not capable to consider a variable number of traffic participants. Therefore we propose an invariant environment representation from the perspective of the ego vehicle. The representation encodes all necessary information for safe decision making. To assess the generalization capabilities of the novel environment representation, we train our agents on a small subset of scenarios and evaluate on the entire diverse set of scenarios. Here we show that the agents are capable to generalize successfully to unseen scenarios, due to the abstraction. In addition we present a simple occlusion model that enables our agents to navigate intersections with occlusions without a significant change in performance.

Karl Kurzer, Christoph Hörtnagl, J. Marius Zöllner

Monte Carlo Tree Search (MCTS) has proven to be capable of solving challenging tasks in domains such as Go, chess and Atari. Previous research has developed parallel versions of MCTS, exploiting today's multiprocessing architectures. These studies focused on versions of MCTS for the discrete case. Our work builds upon existing parallelization strategies and extends them to continuous domains. In particular, leaf parallelization and root parallelization are studied and two final selection strategies that are required to handle continuous states in root parallelization are proposed. The evaluation of the resulting parallelized continuous MCTS is conducted using a challenging cooperative multi-agent system trajectory planning task in the domain of automated vehicles.

Karl Kurzer, Marcus Fechner, J. Marius Zöllner

Efficient driving in urban traffic scenarios requires foresight. The observation of other traffic participants and the inference of their possible next actions depending on the own action is considered cooperative prediction and planning. Humans are well equipped with the capability to predict the actions of multiple interacting traffic participants and plan accordingly, without the need to directly communicate with others. Prior work has shown that it is possible to achieve effective cooperative planning without the need for explicit communication. However, the search space for cooperative plans is so large that most of the computational budget is spent on exploring the search space in unpromising regions that are far away from the solution. To accelerate the planning process, we combined learned heuristics with a cooperative planning method to guide the search towards regions with promising actions, yielding better solutions at lower computational costs.

Larissa T. Triess, David Peter, Christoph B. Rist et al.

This paper presents a novel CNN-based approach for synthesizing high-resolution LiDAR point cloud data. Our approach generates semantically and perceptually realistic results with guidance from specialized loss-functions. First, we utilize a modified per-point loss that addresses missing LiDAR point measurements. Second, we align the quality of our generated output with real-world sensor data by applying a perceptual loss. In large-scale experiments on real-world datasets, we evaluate both the geometric accuracy and semantic segmentation performance using our generated data vs. ground truth. In a mean opinion score testing we further assess the perceptual quality of our generated point clouds. Our results demonstrate a significant quantitative and qualitative improvement in both geometry and semantics over traditional non CNN-based up-sampling methods.

Michael Weber, Michael Fürst, J. Marius Zöllner

Current state-of-the-art object detection algorithms still suffer the problem of imbalanced distribution of training data over object classes and background. Recent work introduced a new loss function called focal loss to mitigate this problem, but at the cost of an additional hyperparameter. Manually tuning this hyperparameter for each training task is highly time-consuming. With automated focal loss we introduce a new loss function which substitutes this hyperparameter by a parameter that is automatically adapted during the training progress and controls the amount of focusing on hard training examples. We show on the COCO benchmark that this leads to an up to 30% faster training convergence. We further introduced a focal regression loss which on the more challenging task of 3D vehicle detection outperforms other loss functions by up to 1.8 AOS and can be used as a value range independent metric for regression.