arXiv subjects

J. Marius Zöllner

Publications and source records attributed to J. Marius Zöllner.

At least 55 records · Page 3Linked to original sources

arXiv2025

LanePerf: a Performance Estimation Framework for Lane Detection

Lane detection is a critical component of Advanced Driver-Assistance Systems (ADAS) and Automated Driving System (ADS), providing essential spatial information for lateral control. However, domain shifts often undermine model reliability when deployed in new environments. Ensuring the robustness and safety of lane detection models typically requires collecting and annotating target domain data, which is resource-intensive. Estimating model performance without ground-truth labels offers a promising alternative for efficient robustness assessment, yet remains underexplored in lane detection. While previous work has addressed performance estimation in image classification, these methods are not directly applicable to lane detection tasks. This paper first adapts five well-performing performance estimation methods from image classification to lane detection, building a baseline. Addressing the limitations of prior approaches that solely rely on softmax scores or lane features, we further propose a new Lane Performance Estimation Framework (LanePerf), which integrates image and lane features using a pretrained image encoder and a DeepSets-based architecture, effectively handling zero-lane detection scenarios and large domain-shift cases. Extensive experiments on the OpenLane dataset, covering diverse domain shifts (scenes, weather, hours), demonstrate that our LanePerf outperforms all baselines, achieving a lower MAE of 0.117 and a higher Spearman's rank correlation coefficient of 0.727. These findings pave the way for robust, label-free performance estimation in ADAS, supporting more efficient testing and improved safety in challenging driving scenarios.

cs.CV↗

arXiv2025

Why Braking? Scenario Extraction and Reasoning Utilizing LLM

The growing number of ADAS-equipped vehicles has led to a dramatic increase in driving data, yet most of them capture routine driving behavior. Identifying and understanding safety-critical corner cases within this vast dataset remains a significant challenge. Braking events are particularly indicative of potentially hazardous situations, motivating the central question of our research: Why does a vehicle brake? Existing approaches primarily rely on rule-based heuristics to retrieve target scenarios using predefined condition filters. While effective in simple environments such as highways, these methods lack generalization in complex urban settings. In this paper, we propose a novel framework that leverages Large Language Model (LLM) for scenario understanding and reasoning. Our method bridges the gap between low-level numerical signals and natural language descriptions, enabling LLM to interpret and classify driving scenarios. We propose a dual-path scenario retrieval that supports both category-based search for known scenarios and embedding-based retrieval for unknown Out-of-Distribution (OOD) scenarios. To facilitate evaluation, we curate scenario annotations on the Argoverse 2 Sensor Dataset. Experimental results show that our method outperforms rule-based baselines and generalizes well to OOD scenarios.

cs.AI↗

arXiv2025

2.5D Object Detection for Intelligent Roadside Infrastructure

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

cs.CV↗

arXiv2025

Exploring Semantic Clustering and Similarity Search for Heterogeneous Traffic Scenario Graph

Scenario-based testing is an indispensable instrument for the comprehensive validation and verification of automated vehicles (AVs). However, finding a manageable and finite, yet representative subset of scenarios in a scalable, possibly unsupervised manner is notoriously challenging. Our work is meant to constitute a cornerstone to facilitate sample-efficient testing, while still capturing the diversity of relevant operational design domains (ODDs) and accounting for the "long tail" phenomenon in particular. To this end, we first propose an expressive and flexible heterogeneous, spatio-temporal graph model for representing traffic scenarios. Leveraging recent advances of graph neural networks (GNNs), we then propose a self-supervised method to learn a universal embedding space for scenario graphs that enables clustering and similarity search. In particular, we implement contrastive learning alongside a bootstrapping-based approach and evaluate their suitability for partitioning the scenario space. Experiments on the nuPlan dataset confirm the model's ability to capture semantics and thus group related scenarios in a meaningful way despite the absence of discrete class labels. Different scenario types materialize as distinct clusters. Our results demonstrate how variable-length traffic scenarios can be condensed into single vector representations that enable nearest-neighbor retrieval of representative candidates for distinct scenario categories. Notably, this is achieved without manual labeling or bias towards an explicit objective such as criticality. Ultimately, our approach can serve as a basis for scalable selection of scenarios to further enhance the efficiency and robustness of testing AVs in simulation.

cs.LG↗

arXiv2025

DigiT4TAF -- Bridging Physical and Digital Worlds for Future Transportation Systems

In the future, mobility will be strongly shaped by the increasing use of digitalization. Not only will individual road users be highly interconnected, but also the road and associated infrastructure. At that point, a Digital Twin becomes particularly appealing because, unlike a basic simulation, it offers a continuous, bilateral connection linking the real and virtual environments. This paper describes the digital reconstruction used to develop the Digital Twin of the Test Area Autonomous Driving-Baden-Württemberg (TAF-BW), Germany. The TAF-BW offers a variety of different road sections, from high-traffic urban intersections and tunnels to multilane motorways. The test area is equipped with a comprehensive Vehicle-to-Everything (V2X) communication infrastructure and multiple intelligent intersections equipped with camera sensors to facilitate real-time traffic flow monitoring. The generation of authentic data as input for the Digital Twin was achieved by extracting object lists at the intersections. This process was facilitated by the combined utilization of camera images from the intelligent infrastructure and LiDAR sensors mounted on a test vehicle. Using a unified interface, recordings from real-world detections of traffic participants can be resimulated. Additionally, the simulation framework's design and the reconstruction process is discussed. The resulting framework is made publicly available for download and utilization at: https://digit4taf-bw.fzi.de The demonstration uses two case studies to illustrate the application of the digital twin and its interfaces: the analysis of traffic signal systems to optimize traffic flow and the simulation of security-related scenarios in the communications sector.

cs.RO↗

arXiv2025

Fool the Stoplight: Realistic Adversarial Patch Attacks on Traffic Light Detectors

Realistic adversarial attacks on various camera-based perception tasks of autonomous vehicles have been successfully demonstrated so far. However, only a few works considered attacks on traffic light detectors. This work shows how CNNs for traffic light detection can be attacked with printed patches. We propose a threat model, where each instance of a traffic light is attacked with a patch placed under it, and describe a training strategy. We demonstrate successful adversarial patch attacks in universal settings. Our experiments show realistic targeted red-to-green label-flipping attacks and attacks on pictogram classification. Finally, we perform a real-world evaluation with printed patches and demonstrate attacks in the lab settings with a mobile traffic light for construction sites and in a test area with stationary traffic lights. Our code is available at https://github.com/KASTEL-MobilityLab/attacks-on-traffic-light-detection.

cs.CV↗

arXiv2025

Contrast & Compress: Learning Lightweight Embeddings for Short Trajectories

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.

cs.CV↗

arXiv2025

Functionality Assessment Framework for Autonomous Driving Systems using Subjective Networks

In complex autonomous driving (AD) software systems, the functioning of each system part is crucial for safe operation. By measuring the current functionality or operability of individual components an isolated glimpse into the system is given. Literature provides several of these detached assessments, often in the form of safety or performance measures. But dependencies, redundancies, error propagation and conflicting functionality statements do not allow for easy combination of these measures into a big picture of the functioning of the entire AD stack. Data is processed and exchanged between different components, each of which can fail, making an overall statement challenging. The lack of functionality assessment frameworks that tackle these problems underlines this complexity. This article presents a novel framework for inferring an overall functionality statement for complex component based systems by considering their dependencies, redundancies, error propagation paths and the assessments of individual components. Our framework first incorporates a comprehensive conversion to an assessment representation of the system. The representation is based on Subjective Networks (SNs) that allow for easy identification of faulty system parts. Second, the framework offers a flexible method for computing the system's functionality while dealing with contradicting assessments about the same component and dependencies, as well as redundancies, of the system. We discuss the framework's capabilities on real-life data of our AD stack with assessments of various components.

cs.RO↗

arXiv2025

Online Performance Assessment of Multi-Source-Localization for Autonomous Driving Systems Using Subjective Logic

Autonomous driving (AD) relies heavily on high precision localization as a crucial part of all driving related software components. The precise positioning is necessary for the utilization of high-definition maps, prediction of other road participants and the controlling of the vehicle itself. Due to this reason, the localization is absolutely safety relevant. Typical errors of the localization systems, which are long term drifts, jumps and false localization, that must be detected to enhance safety. An online assessment and evaluation of the current localization performance is a challenging task, which is usually done by Kalman filtering for single localization systems. Current autonomous vehicles cope with these challenges by fusing multiple individual localization methods into an overall state estimation. Such approaches need expert knowledge for a competitive performance in challenging environments. This expert knowledge is based on the trust and the prioritization of distinct localization methods in respect to the current situation and environment. This work presents a novel online performance assessment technique of multiple localization systems by using subjective logic (SL). In our research vehicles, three different systems for localization are available, namely odometry-, Simultaneous Localization And Mapping (SLAM)- and Global Navigation Satellite System (GNSS)-based. Our performance assessment models the behavior of these three localization systems individually and puts them into reference of each other. The experiments were carried out using the CoCar NextGen, which is based on an Audi A6. The vehicle's localization system was evaluated under challenging conditions, specifically within a tunnel environment. The overall evaluation shows the feasibility of our approach.

cs.RO↗

arXiv2025

Centralized Decision-Making for Platooning By Using SPaT-Driven Reference Speeds

This paper introduces a centralized approach for fuel-efficient urban platooning by leveraging real-time Vehicle- to-Everything (V2X) communication and Signal Phase and Timing (SPaT) data. A nonlinear Model Predictive Control (MPC) algorithm optimizes the trajectories of platoon leader vehicles, employing an asymmetric cost function to minimize fuel-intensive acceleration. Following vehicles utilize a gap- and velocity-based control strategy, complemented by dynamic platoon splitting logic communicated through Platoon Control Messages (PCM) and Platoon Awareness Messages (PAM). Simulation results obtained from the CARLA environment demonstrate substantial fuel savings of up to 41.2%, along with smoother traffic flows, fewer vehicle stops, and improved intersection throughput.

cs.RO↗

arXiv2025

Hybrid Video Anomaly Detection for Anomalous Scenarios in Autonomous Driving

In autonomous driving, the most challenging scenarios can only be detected within their temporal context. Most video anomaly detection approaches focus either on surveillance or traffic accidents, which are only a subfield of autonomous driving. We present HF$^2$-VAD$_{AD}$, a variation of the HF$^2$-VAD surveillance video anomaly detection method for autonomous driving. We learn a representation of normality from a vehicle's ego perspective and evaluate pixel-wise anomaly detections in rare and critical scenarios.

cs.CV↗

arXiv2025

The ATLAS of Traffic Lights: A Reliable Perception Framework for Autonomous Driving

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.

cs.CV↗

arXiv2025

MUVO: A Multimodal Generative World Model for Autonomous Driving with Geometric Representations

World models for autonomous driving have the potential to dramatically improve the reasoning capabilities of today's systems. However, most works focus on camera data, with only a few that leverage lidar data or combine both to better represent autonomous vehicle sensor setups. In addition, raw sensor predictions are less actionable than 3D occupancy predictions, but there are no works examining the effects of combining both multimodal sensor data and 3D occupancy prediction. In this work, we perform a set of experiments with a MUltimodal World Model with Geometric VOxel representations (MUVO) to evaluate different sensor fusion strategies to better understand the effects on sensor data prediction. We also analyze potential weaknesses of current sensor fusion approaches and examine the benefits of additionally predicting 3D occupancy.

cs.LG↗

arXiv2025

Label-Free Model Failure Detection for Lidar-based Point Cloud Segmentation

Autonomous vehicles drive millions of miles on the road each year. Under such circumstances, deployed machine learning models are prone to failure both in seemingly normal situations and in the presence of outliers. However, in the training phase, they are only evaluated on small validation and test sets, which are unable to reveal model failures due to their limited scenario coverage. While it is difficult and expensive to acquire large and representative labeled datasets for evaluation, large-scale unlabeled datasets are typically available. In this work, we introduce label-free model failure detection for lidar-based point cloud segmentation, taking advantage of the abundance of unlabeled data available. We leverage different data characteristics by training a supervised and self-supervised stream for the same task to detect failure modes. We perform a large-scale qualitative analysis and present LidarCODA, the first publicly available dataset with labeled anomalies in real-world lidar data, for an extensive quantitative analysis.

cs.RO↗

arXiv2025

A Chefs KISS -- Utilizing semantic information in both ICP and SLAM framework

For utilizing autonomous vehicle in urban areas a reliable localization is needed. Especially when HD maps are used, a precise and repeatable method has to be chosen. Therefore accurate map generation but also re-localization against these maps is necessary. Due to best 3D reconstruction of the surrounding, LiDAR has become a reliable modality for localization. The latest LiDAR odometry estimation are based on iterative closest point (ICP) approaches, namely KISS-ICP and SAGE-ICP. We extend the capabilities of KISS-ICP by incorporating semantic information into the point alignment process using a generalizable approach with minimal parameter tuning. This enhancement allows us to surpass KISS-ICP in terms of absolute trajectory error (ATE), the primary metric for map accuracy. Additionally, we improve the Cartographer mapping framework to handle semantic information. Cartographer facilitates loop closure detection over larger areas, mitigating odometry drift and further enhancing ATE accuracy. By integrating semantic information into the mapping process, we enable the filtering of specific classes, such as parked vehicles, from the resulting map. This filtering improves relocalization quality by addressing temporal changes, such as vehicles being moved.

cs.RO↗

arXiv2025

Self-Supervised Pretraining for Aerial Road Extraction

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.

cs.CV↗

arXiv2025

From One to the Power of Many: Invariance to Multi-LiDAR Perception from Single-Sensor Datasets

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.

cs.CV↗

arXiv2024

Towards Adversarial Robustness of Model-Level Mixture-of-Experts Architectures for Semantic Segmentation

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/}.

cs.CV↗