Keywords: Lidar-Based Environment Mapping
Abstract: Future automated vehicles are likely to rely on LiDAR sensors to perform automated functions. These LiDAR sensors have the potential to interfere with each other, with this mutual interference being represented by unwanted points in the point cloud. These points could affect the quality of perception of objects, such as potentially creating false objects or obscuring real objects. Hence, it is necessary to identify interference points and remove them from the point cloud, without removing points that belong to valid objects. This paper presents a novel LiDAR mutual interference mitigation algorithm that successfully mitigates interference and potentially can be performed in real time. The number of interference points when the victim and offending LiDARs were separated by 1m was reduced from over 1000 points to under 40 points. The proposed method performed better than the current state-of-the-art techniques on most metrics and was designed to ensure that consecutive points at the same range, such as those belonging to small objects at far distances, were not permanently removed from consecutive point clouds. These results ensure a removal of interference point without affecting the accuracy of perception tasks based on LiDAR data.

Keywords: Human-Machine Interface (HMI) Design Principles, Feedback Systems for Driver Interaction, Human Factors Analysis in Vehicle Design
Abstract: To mitigate the occurrence of traffic accidents, addressing the human factors pertinent to road safety, particularly driver operational errors, is essential. Drivers in urban environments can benefit from readiness evaluations and feedback, which help mitigate errors, promote safer driving speeds, and enhance situational awareness. This study aimed to develop and validate a driver behavior assessment system centered around the concept of stepwise driver readiness. The system is designed to evaluate driver behavior and provide progressive feedback to improve performance. We extrapolated the original concept of driver readiness to create a stepwise readiness evaluation system, which was implemented and tested on urban roads to evaluate its effectiveness. The results demonstrate that the system effectively evaluates readiness, and the stepwise feedback mechanism significantly enhances driver performance. Notably, the success of the feedback process was influenced by the level of driver acceptance. These results highlight the importance of the expanded driver readiness concept in managing human factor-related driving risks and improving road safety.

Keywords: Synthetic Data Generation for Training, 3D Scene Reconstruction Methods, Vulnerable Road User Protection Strategies
Abstract: In Autonomous Driving (AD) Perception, cyclists are considered safety-critical scene objects. Commonly used publicly-available AD datasets typically contain large amounts of car and vehicle object instances but a low number of cyclist instances, usually with limited appearance and pose diversity. This cyclist training data scarcity problem not only limits the generalization of deep-learning perception models for cyclist semantic segmentation, pose estimation, and cyclist crossing intention prediction, but also limits research on new cyclist-related tasks such as fine-grained cyclist pose estimation and spatio-temporal analysis under complex interactions between humans and articulated objects. To address this data scarcity problem, in this paper we propose a framework to generate synthetic dynamic 3D cyclist data assets that can be used to generate training data for different tasks. In our framework, we designed a methodology for creating a new part-based multi-view articulated synthetic 3D bicycle dataset that we call 3DArticBikes that we use to train a 3D Gaussian Splatting (3DGS)-based reconstruction and image rendering method. We then propose a parametric bicycle 3DGS composition model to assemble 8-DoF pose-controllable 3D bicycles. Finally, using dynamic information from cyclist videos, we build a complete synthetic dynamic 3D cyclist (rider pedaling a bicycle) by re-posing a selectable synthetic 3D person, while automatically placing the rider onto one of our new articulated 3D bicycles using a proposed 3D Keypoint optimization-based Inverse Kinematics pose refinement. We present both, qualitative and quantitative results where we compare our generated cyclists against those from a recent stable diffusion-based method.

Keywords: Application of Neural Fields in Autonomous Driving, Scalable Neural Scene Representation, 3D Scene Reconstruction Methods
Abstract: Current 3D reconstruction techniques struggle to infer unbounded scenes from a few images faithfully. Most existing methods have high computational demands, require detailed pose information, and cannot reconstruct occluded regions reliably. We introduce 6Img-to-3D, a novel transformer-based encoder-renderer method for single-shot image-to-3D reconstruction. Our method outputs a 3D-consistent parameterized triplane from only six outward-facing input images for large-scale, unbounded outdoor driving scenarios. We take a step towards resolving existing shortcomings by combining contracted custom cross- and self-attention mechanisms for triplane parameterization, differentiable volume rendering, scene contraction, and image feature projection. We showcase on synthetic data that six surround-view vehicle images from a single timestamp are enough to reconstruct 360^{circ} scenes during inference time, taking 395 ms. Our method allows, for example, rendering third-person images and birds-eye views. Code, and more results are available at https://6Img-to-3D.GitHub.io/.

Keywords: Teleoperation Control Systems for Vehicles, Application of Neural Fields in Autonomous Driving
Abstract: In this paper, we propose models with explicit trainable delays for learning teleoperated driving (ToD) behavior from limited vehicle trajectory data. The data-driven model is integrated with physics-based nonlinear vehicle dynamics and formulated as a neural delay differential equation (NDDE). The model can be analyzed using the same tools as developed for classical delay differential equations. The physics-based nonlinearity built into the data-driven model reduces the model complexity, enables training with limited data, and provides good generalizations. An overall latency in the loop is learned and a generic steering controller that characterizes the remote operator is identified at the same time through the training process. This information could be used to evaluate the performance of ToD in the presence of communication latency. We provide examples of learning from simulation data generated by a kinematic vehicle model and from experimental data generated by a human operator driving in a high-fidelity simulation environment. The same data-driven model and training algorithm is used in both cases, which demonstrates the generalizability of the proposed approach.

Keywords: 3D Scene Reconstruction Methods, Geometric vs. Semantic Mapping, UAV Datasets
Abstract: Road surface reconstruction from aerial images is fundamental for autonomous driving, urban planning, and virtual simulation, where smoothness, compactness, and accuracy are critical quality factors. Existing reconstruction methods often produce artifacts and inconsistencies that limit usability, while downstream tasks have a tendency to represent roads as planes for simplicity but at the cost of accuracy. We introduce FlexRoad, the first framework to directly address road surface smoothing by fitting Non-Uniform Rational B-Splines (NURBS) surfaces to 3D road points obtained from photogrammetric reconstructions or geodata providers. Our method at its core utilizes the Elevation-Constrained Spatial Road Clustering (ECSRC) algorithm for robust anomaly correction, significantly reducing surface roughness and fitting errors. To facilitate quantitative comparison between road surface reconstruction methods, we present GeoRoad Dataset (GeRoD), a diverse collection of road surface and terrain profiles derived from openly accessible geodata. Experiments on GeRoD and the photogrammetry-based DeepScenario Open 3D Dataset (DSC3D) demonstrate that FlexRoad considerably surpasses commonly used road surface representations across various metrics while being insensitive to various input sources, terrains, and noise types. By performing ablation studies, we identify the key role of each component towards high-quality reconstruction performance, making FlexRoad a generic method for realistic road surface modeling.

Keywords: Perception Algorithms for Adverse Weather Conditions, Semantic Segmentation Techniques, Instance and Panoptic Segmentation Techniques
Abstract: Enhancement of images from RGB cameras is of particular interest due to its wide range of ever-increasing applications such as medical imaging, satellite imaging, automated driving, etc. In autonomous driving, various techniques are used to enhance image quality under challenging lighting conditions. These include artificial augmentation to improve visibility in poor nighttime conditions, illumination-invariant imaging to reduce the impact of lighting variations, and shadow mitigation to ensure consistent image clarity in bright daylight. This paper proposes a pipeline for Shadow Erosion and Nighttime Adaptability in images for automated driving applications while preserving color and texture details. The Shadow Erosion and Nighttime Adaptability pipeline is compared to the widely used CLAHE technique and evaluated based on illumination uniformity and visual perception quality metrics. The results also demonstrate a significant improvement over CLAHE, enhancing a YOLO-based drivable area segmentation algorithm.

Keywords: Automotive Datasets, Perception Algorithms for Adverse Weather Conditions
Abstract: The problem of calibration from straight lines is fundamental in geometric computer vision, with wellestablished theoretical foundations. However, its practical applicability remains limited, particularly in real-world outdoor scenarios. These environments pose significant challenges due to diverse and cluttered scenes, interrupted reprojections of straight 3D lines, and varying lighting conditions, making the task notoriously difficult. Furthermore, the field lacks a dedicated dataset encouraging the development of respective detection algorithms. In this study, we present a small dataset named "ClearLines", and by detailing its creation process, provide practical insights that can serve as a guide for developing and refining straight 3D line detection algorithms.

Keywords: Deep Learning Based Approaches, Collision Avoidance Algorithms, Smart City Mobility Integration Strategies
Abstract: In complex transportation systems, accurately sensing the surrounding environment and predicting the risk of potential accidents is crucial. Most existing accident prediction methods are based on temporal neural networks, such as RNN and LSTM. Recent multimodal fusion approaches improve vehicle localization through 3D target detection and assess potential risks by calculating inter-vehicle distances. However, these temporal networks and multimodal fusion methods suffer from limited detection robustness and high economic costs. To address these challenges, we propose AccidentBlip, a vision-only framework that employs our self-designed Motion Accident Transformer (MA-former) to process each frame of video. Unlike conventional self-attention mechanisms, MA-former replaces Q-former's self-attention with temporal attention, allowing the query corresponding to the previous frame to generate the query input for the next frame. Additionally, we introduce a residual module connection between queries of consecutive frames to enhance the model's temporal processing capabilities. For complex V2V and V2X scenarios, AccidentBlip adapts by concatenating queries from multiple cameras, effectively capturing spatial and temporal relationships. In particular, AccidentBlip achieves SOTA performance in both accident detection and prediction tasks on the DeepAccident dataset. It also outperforms current SOTA methods in V2V and V2X scenarios, demonstrating a superior capability to understand complex real-world environments.

Keywords: Deep Learning Based Approaches, Semantic Segmentation Techniques
Abstract: GNSS as the input for train localization, testing GNSS positioning for train localization as function and performance is a necessary procedure throughout the entire lifecycle of the train control system, from design to operation. This paper focuses on analyzing the GNSS data recorded during train operation, considering both textual records and numerical data generated by GNSS receivers. The textual records come from regular records logged by the trainborne equipment of ITCS collected on the Qinghai-Tibet railway line, while the numerical data is extracted from the processed NMEA data and RINEX format files. A semantic analysis method based on the TF-IDF algorithm is used to mine the textual records, identifying important feature words and utilizing fault tree analysis to determine the causes of the failure modes. For the numerical data, a machine learning approach based on SSA-XGBoost is employed, followed by the application of the Shapley additive explanation method to identify the most important parameters. We identified that Bad satellite geometry" Poor satellite signal quality" and Insufficient Number of visible satellites " are important fault causes when compiling test cases. Through the mining and analysis of numerical data, we have determined that it is crucial to model the 3 key parameters of SNR_mean, HDOP, and PRerror_mean during the testing process. By mining and analyzing these textual records and numerical data, the paper provides clear directions and effective foundations for the generation of GNSS test cases for train localization. Based on the mining results, key failure modes and parameters to design more targeted and comprehensive test cases, improving testing effectiveness and comprehensiveness.

Keywords: Dataset Augmentation Using Neural Field, Application of Neural Fields in Autonomous Driving, Techniques for Dataset Domain Adaptation
Abstract: Online mapping reduces the reliance of autonomous vehicles on high-definition (HD) maps, significantly enhancing scalability. However, recent advancements often overlook cross-sensor configuration generalization, leading to performance degradation when models are deployed on vehicles with different camera intrinsics and extrinsics. With the rapid evolution of novel view synthesis methods, we investigate the extent to which these techniques can be leveraged to address the sensor configuration generalization challenge. We propose a novel framework leveraging Gaussian splatting to reconstruct scenes and render camera images in target sensor configurations. The target config sensor data, along with labels mapped to the target config, are used to train online mapping models. Our proposed framework on the nuScenes and Argoverse 2 datasets demonstrates a performance improvement of 18% through effective dataset augmentation, achieves faster convergence and efficient training, and exceeds state-of-the-art performance when using only 25% of the original training data. This enables data reuse and reduces the need for laborious data labeling. Project page at https://henryzhangzhy.github.io/mapgs.

Keywords: Application of Neural Fields in Autonomous Driving, Integrating Diverse Data Sources (e.g HD maps, LIDAR) in Neural Scene Representations, 3D Scene Reconstruction Methods
Abstract: In this paper, we propose a Neural Radiance Fields (NeRF) based framework, referred to as Novel View Synthesis Framework (NVSF). It jointly learns the implicit neural representation of space and time-varying scene for both LiDAR and Camera. We test this on a real-world autonomous driving scenario containing both static and dynamic scenes. Compared to existing multimodal dynamic NeRFs, our framework is selfsupervised, thus eliminating the need for 3D labels. For efficient training and faster convergence, we introduce heuristic based image pixel sampling to focus on pixels with rich information. To preserve the local features of LiDAR points, a Double Gradient based mask is employed. Extensive experiments on the KITTI-360 dataset show that, compared to the baseline models, our Framework has reported best performance on both LiDAR and Camera domain. Code of the model is available at https://github.com/gaurav00700/Selfsupervised-NVSF

Keywords: High-Resolution Image Sensing Techniques, Level 3 Driving Systems Architecture and Techniques
Abstract: Existing wired communication vehicle network technology lacks the bandwidth required to support the data rates produced by the sensor suite for assisted and automated driving functions in the next generation of intelligent vehicles. Video from image sensors demands a very high bandwidth and cameras are continually being developed giving better resolution, bit-depth (dynamic range) and frame rate. To overcome this challenge, compression is a potential solution to reduce the required bandwidth. As modern automotive cameras produce Bayer images, it might be more effective to compress Bayer images directly instead of red-green-blue (RGB) images compression, which is currently widely deployed in compression schemes. By using Bayer, we avoid to triple the memory storage and preserve data fidelity by bypassing demosaicing and other non-reversible processes in the Image Signal Processing (ISP) pipeline. Bayer adaption techniques indicate methods to convert Bayer data to be used in traditional compression pipelines. In this research, we propose two novel Colour Space Transform (CST) techniques and implement them in combination with the widely used H.264 codec. The presented experiments show that the novel CST techniques are superior to direct and Separation Bayer adaption techniques when paired with object detection, specifically when the bit rate is between ≈750 to 1250 kb/fr. These results can support the future adoption of Bayer data and compression in future vehicles, addressing the problem of unsustainable perception sensors’ datarates.

Keywords: Techniques for Dataset Domain Adaptation, Synthetic Data Generation for Training, Deep Learning Based Approaches
Abstract: In the field of autonomous driving, simulators have gained growing attention as they provide a cost-effective environment to collect data with various sensor modalities across diverse driving scenarios. Yet, the domain gap between real and synthetic data remains a significant barrier. To address this, numerous methods have been proposed to enhance the usability of synthetic data by bridging the gap. However, many of those approaches often focus on a single sensor setup, either a monocular camera or a LiDAR, potentially leading to inconsistent domain adaptation across sensors. As recent perception models actively employ sensor-fusion approaches like camera+LiDAR, we propose Sensor-wise Offline Dataset Adaptation (SODA), designed for a multimodal synthetic driving dataset with multiview cameras and a LiDAR. It ensures consistency between sensor modalities after domain adaptation by enhancing signal information (camera pixel and LiDAR intensity values) while preserving their shape (image contents, point cloud geometry). To validate our method, MORDA (synthetic) and nuScenes (real) are opted as source and target datasets. In our extensive evaluation, incorporating SODA-applied MORDA during training of BEVFusion (camera+LiDAR) achieves mAP gain of 1.5% compared to training the same model solely on nuScenes. Further, ~2% improvement in mAP is observed under challenging adverse condition scenarios.

Keywords: Foundation Models Based Approaches, Level 4-5 Autonomous Driving Systems Architecture, Semantic Segmentation Techniques
Abstract: This paper introduces an innovative application of foundation models, enabling Unmanned Ground Vehicles (UGVs) equipped with an RGB-D camera to navigate to designated destinations based on human language instructions. Unlike learning-based methods, this approach does not require prior training but instead leverages existing foundation models, thus facilitating generalization to novel environments. Upon receiving human language instructions, these are transformed into a `cognitive route description' using a large language model (LLM)—a detailed navigation route expressed in human language. The vehicle then decomposes this description into landmarks and navigation maneuvers. The vehicle also determines elevation costs and identifies navigability levels of different regions through a terrain segmentation model, GANav, trained on open datasets. Semantic elevation costs, which take both elevation and navigability levels into account, are estimated and provided to the Model Predictive Path Integral (MPPI) planner, responsible for local path planning. Concurrently, the vehicle searches for target landmarks using foundation models, including YOLO-World and EfficientViT-SAM. Ultimately, the vehicle executes the navigation commands to reach the designated destination, the final landmark. Our experiments demonstrate that this application successfully guides UGVs to their destinations following human language instructions in novel environments, such as unfamiliar terrain or urban settings.

Keywords: Synthetic Data Generation for Training, Deep Learning Based Approaches, Automotive Datasets
Abstract: In the high-resolution image style transfer task of road traffic scenes, how to fully transfer style information while retaining the original content structure information is a challenging problem. In this paper, a novel Transformer-based generative adversarial network for high-resolution unpaired image translation is proposed. Firstly, we design a style transformation module based on cross adaptive Transformer, which dynamically adjusts the content features to achieve statistical alignment between content features and the target style. Meanwhile, an image frequency-domain enhancement module is designed based on cross-attention, which fuses the global information of the low-frequency style with the local details of the high-frequency content information. The detailed texture is then enhanced while the image style consistency is maintained. Furthermore, we design a threshold-guided negative sample screening strategy based on contrastive learning, which can improves the model's transfer effect. The experimental results well demonstrate the effectiveness of the proposed method.

Keywords: Static and Dynamic Object Detection Algorithms, Automotive Datasets, Integrating HD Maps and Perception Data in Neural Networks
Abstract: 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 integration 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.

Keywords: Deep Learning Based Approaches, Representation Learning for Driving Scenarios
Abstract: This paper investigates the integration of graph neural networks (GNNs) with Qualitative Explainable Graphs (QXGs) for scene understanding in automated driving. Scene understanding is the basis for any further reactive or proactive decision-making. Scene understanding and related reasoning is inherently an explanation task: why is another traffic participant doing something, what or who caused their actions? While previous work demonstrated QXGs' effectiveness using shallow machine learning models, these approaches were limited to analysing single relation chains between object pairs, disregarding the broader scene context. We propose a novel GNN architecture that processes entire graph structures to identify relevant objects in traffic scenes. We evaluate our method on the nuScenes dataset enriched with DriveLM's human-annotated relevance labels. Experimental results show that our GNN-based approach achieves superior performance compared to baseline methods. The model effectively handles the inherent class imbalance in relevant object identification tasks while considering the complete spatial-temporal relationships between all objects in the scene. Our work demonstrates the potential of combining qualitative representations with deep learning approaches for explainable scene understanding in autonomous driving systems.

Keywords: Electric and Hybrid Vehicle Integration, Data Augmentation Techniques Using Neural Networks, Deep Learning Based Approaches
Abstract: The growing adoption of electric vehicles (EVs) has heightened the need for reliable and efficient On-Board Chargers (OBCs). Power Factor Correction (PFC) circuits within OBCs are critical for optimizing energy conversion and minimizing power losses. However, fault diagnosis in PFC circuits remains a challenge due to the difficulty of replicating real-world fault scenarios for data collection. This study addresses these challenges by employing Generative Adversarial Networks (GANs) to augment fault signal data. By generating diverse and realistic fault signals, this approach enhances the robustness of fault classification models. The proposed CRNNWGAN model, a fusion of C-RNN-GAN and WGAN-GP, effectively captures temporal dependencies and improves the accuracy of fault diagnosis. Experimental results demonstrate the superiority of the augmented dataset in classification tasks, providing a scalable solution for improving the reliability of EV charging systems.

Keywords: Foundation Models Based Approaches, Deep Learning Based Approaches, Representation Learning for Driving Scenarios
Abstract: Multimodal large language models (MLLMs) have shown significant potential in enhancing driving scene understanding and visual question answering (VQA) through advanced logical reasoning capabilities. These tasks support driving action generation and explanation, especially in end-to-end autonomous driving applications. However, deploying these models poses a significant challenge due to their substantial parameter sizes and computational demands, which often exceed onboard computational limits. A key limitation stems from the large number of visual tokens needed to capture detailed, long-context visual information, resulting in increased latency and memory use. To address this, we propose Video Token Sparsification (VTS), a novel approach that leverages redundancy in consecutive video frames to reduce visual tokens while preserving critical information. VTS employs a lightweight CNN-based model to identify key frames and prune less informative tokens, mitigating hallucinations and boosting inference throughput without performance loss. Comprehensive experiments on the LingoQA and DRAMA benchmarks show that VTS achieves up to a 33% improvement in inference throughput and a 28% reduction in memory usage compared to baselines, maintaining comparable performance.

Keywords: Advanced Multisensory Data Fusion Algorithms, Deep Learning Based Approaches, Foundation Models Based Approaches
Abstract: 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.

Keywords: Techniques for Dataset Domain Adaptation, Semantic Segmentation Techniques, Foundation Models Based Approaches
Abstract: Training large-scale vision models for autonomous driving is computationally expensive and requires extensive manual annotation. While reducing dataset size could address these limitations, it typically results in degraded model performance. In this paper, we propose a novel self-supervised data selection framework that leverages vision foundation models to identify and retain high-value training samples, enabling efficient dataset curation without compromising performance. Our approach fine-tunes a foundation model’s vision encoder using a contrastive objective, then perform density-based clustering in its learned embedding space to retain only those samples that maximally preserve semantic diversity. Through experiments, we show that training on our curated subset outperforms models trained on the full dataset, and exceeds random selection in semantic segmentation tasks. Additionally, our comparisons across different foundation model architectures and segmentation backbones provide insights into effective dataset curation. Our results highlight that self-supervised data selection can significantly reduce both annotation and computational overhead, providing a scalable alternative to naively expanding datasets.

Keywords: Reinforcement Learning for Planning, Motion Planning Algorithms for Autonomous Vehicles, Multi-Objective Planning Approaches
Abstract: Autonomous Vehicle (AV) driving involves complex maneuvers, with Obstacle Avoidance (OA) being one of the most challenging and safety critical tasks. While Reinforcement Learning (RL) has shown promise in achieving human like driving behavior, a single RL agent struggles to manage the multiple sub-maneuvers required for OA, particularly in safety critical scenarios. To address this, we propose an Option Policy inspired Hierarchical Deep Reinforcement Learning (OPDRL) framework that divides OA into sub tasks such as left lane change, straight driving, and right lane change. Each sub task is handled by a specialized RL agent, governed by a central master policy. This approach reduces training time, simplifies validation, and seamlessly incorporates traffic safety rules to ensure robust decision making in safety critical scenarios. The proposed method is validated using scenarios inspired by the National Highway Traffic Safety Administration (NHTSA) pre-crash scenarios in the CARLA simulator, demonstrating its effectiveness in handling OA maneuvers efficiently.

Keywords: Motion Planning Algorithms for Autonomous Vehicles, Collision Avoidance Algorithms
Abstract: Trajectory planning for autonomous driving requires safe, robust and efficient algorithms in complex, dynamic environments where multiple vehicles interact. Constrained Dynamic Games (CDG) provide a unifying framework to model such multi-agent interactions, with solutions corresponding to Generalized Nash Equilibria (GNE). Existing solving methods of CDG can be divided into the open-loop strategies, which is computational-efficient but lack robustness to disturbances, and the feedback strategies, which offer adaptability but often struggle to guarantee real-time performance when handling hard constraints. To address these limitations, we introduce ADMM-iCLQG, a solver of CDG tailored for planning a feedback trajectory in real-time. We iteratively approximate the nonlinear CDG with a linearly Constrained Linear Quadratic Game (CLQG). By leveraging the Alternating Direction Method of Multipliers (ADMM), the solver efficiently manages constraints while computing feedback strategies that converge to GNE. We demonstrate the algorithm effectiveness through extensive simulations, showing its superiority in computational efficiency and reliability. Furthermore, physical experiments confirm the approach’s potential of real-time implementation, achieving a planning frequency exceeding 40 Hz for up to six vehicles.

Keywords: End-to-End Neural Network Architectures and Techniques, Reinforcement Learning for Planning, Adaptive Vehicle Control Techniques
Abstract: Parking in narrow vertical parking spaces presents significant challenges for Autonomous Parking Assistance (APA) systems. To address this challenge, a Latent Memory-based Soft Actor-Critic (LM-SAC) algorithm has been presented which leverages latent memory to implicitly store contextual information during the parking process, enabling the agent to precisely maneuver in confined spaces. Meanwhile, a Progressive Milestone Training (PMT) method was proposed that decomposes the task into a series of intermediate subtasks and integrates curriculum learning to enhance training efficiency. Additionally, a virtual parking environment tailored for narrow spaces was developed, featuring real-sized modeling and high-precision collision detection, with vehicle-mounted Virtual-LiDAR Detection (VLD) to represent surrounding obstacle information. Experimental results demonstrate the efficacy of proposed approach, which achieves a parking success rate of 90%, with an average horizontal error of 1.26 cm and a heading error of 0.38 degrees in challenging vertical parking scenarios, where the opposing space is 4.5m wide and the vehicle length is 5.25m.

Keywords: Predictive Trajectory Models and Motion Forecasting, Motion Planning Algorithms for Autonomous Vehicles, Semantic Understanding and Decision-Making with Neural Fields
Abstract: With the increasing demand for handling complex scenarios in autonomous driving, data-driven planning methods based on imitation learning have attracted significant attention. In this context, this work proposes the PlanVMN method, aiming to enhance the model's robustness to data and its causal inference ability in planning. Based on vectorized scene inputs, this method integrates the world model into the core architecture of the planning model and deduces the temporal features of the scene based on the actions of the ego vehicle. Meanwhile, we introduce an attention-based history enhancement component within the world model, which remarkably improves the planning performance of the ego vehicle. Experiments show that the model has achieved outstanding results in the open-loop and closed-loop tests of nuPlan. Notably, thanks to the generative ability endowed by the unique world model architecture, the model can still exhibit excellent performance when facing blind area, sensor misfuctioning or detection instability, providing strong support for planning in complex scenarios of autonomous driving.

Keywords: Real-Time Control Strategies
Abstract: This work develops a robust nonlinear Model Predictive Control (MPC) framework for path tracking in autonomous vehicles operating at the limits of handling utilizing a Control Contraction Metric (CCM) derived from a perturbed dynamic single track model. We first present a nonlinear MPC scheme for autonomous vehicles. Building on this nominal scheme, we assume limited uncertainty in tire parameters as well as bounded force disturbances in both lateral and longitudinal directions. By simplifying the perturbed model, we optimize a CCM for the uncertain model, which is validated through simulations at the dynamic limits of vehicle performance. This CCM is subsequently employed to parameterize a homothetic tube used for constraint tightening within the MPC formulation. The resulting robust nonlinear MPC is computationally more efficient than competing methods, as it introduces only a single additional state variable into the prediction model compared to the nominal scheme. Simulation results demonstrate that the homothetic tube expands most significantly in regions where the nominal scheme would otherwise violate constraints, illustrating its ability to capture all uncertain trajectories while avoiding unnecessary conservatism.

Keywords: Reinforcement Learning for Planning, End-to-End Neural Network Architectures and Techniques, Motion Planning Algorithms for Autonomous Vehicles
Abstract: This paper addresses the challenges of guiding Autonomous Vehicles (AVs) through unprotected left turns at unsignalized, fully uncontrolled, chaotic intersections consisting of pedestrians and adversarial vehicles. These intersections present several challenges for AVs, including the absence of traffic signals or signs, the dynamic nature of the environment, including human interactions due to the presence of crosswalks, and the variability of intersection layouts. This study develops a containerized simulation environment based on a high-fidelity simulation tool, CARLA, and is made open-source at https://github.com/faizansana/intersection-carla-gym. This study proposes using a hierarchical Deep Reinforcement Learning (DRL) wherein the DRL policy sets the target speed for the low-level PID controllers used for actuation. To the best of our knowledge, this study stands as the pioneering effort in addressing pedestrian interaction within a high-fidelity simulation environment, coexisting alongside adversarial vehicles within CARLA. It conducts a comprehensive comparative analysis of five model-free DRL algorithms in a hierarchical framework. It is observed that Recurrent PPO with a discretized action space has the highest success rate and lowest accident rate for unprotected left turns. However, all algorithms struggle to achieve acceptable success rates when pedestrians exist within the environment, emphasizing the importance of the high-fidelity simulation tool. The paper concludes by discussing potential extensions of the proposed system and future research directions.

Keywords: Real-Time Control Strategies, Level 2 ADAS Control Techniques, Motion Planning Algorithms for Autonomous Vehicles
Abstract: This paper proposes a novel physics-informed machine learning framework for motion planning and control of autonomous vehicles. By integrating longitudinal and lateral control, a nonlinear control problem is formulated using Model Predictive Control (MPC). To address computational challenges, a self-supervised framework, Recurrent Predictive Control (RPC), is introduced, leveraging differentiable neural networks and recurrent neural networks to train a neural network controller. Additionally, a heuristic feedback control layer is designed to reduce steady-state errors in the closed-loop tracking. Through numerical simulations and co-simulations using Simulink and CarSim, five neural network controllers are compared with an MPC controller in a lane-changing scenario. The proposed RPC framework improves computational efficiency by 95% compared to MPC, enhances generalization performance compared to Approximate MPC, and reduces performance loss by 17% compared to Differentiable Predictive Control. The heuristic feedback control layer further reduces steady-state errors and improves convergence speed during training.

Keywords: Motion Forecasting, Motion Planning Algorithms for Autonomous Vehicles, Vehicle-to-Infrastructure (V2I) Communication
Abstract: Accurate and real-time prediction of surrounding vehicles’ lane-changing intentions is a critical challenge in deploying safe and efficient autonomous driving systems in open-world scenarios. Existing high-performing methods remain hard to deploy due to their high computational cost, long training times, and excessive memory requirements. Here, we propose an efficient lane-changing intention prediction approach based on brain-inspired Spiking Neural Networks (SNN). By leveraging the event-driven nature of SNN, the proposed approach enables us to encode the vehicle's states in a more efficient manner. Comparison experiments conducted on HighD and NGSIM datasets demonstrate that our method significantly improves training efficiency and reduces deployment costs while maintaining comparable prediction accuracy. Particularly, compared to the baseline, our approach reduces training time by 75% and memory usage by 99.9%. These results validate the efficiency and reliability of our method in lane-changing predictions, highlighting its potential for safe and efficient autonomous driving systems while offering significant advantages in deployment, including reduced training time, lower memory usage, and faster inference.

Keywords: Adaptive Vehicle Control Techniques, Level 2 ADAS Control Techniques, End-to-End Neural Network Architectures and Techniques
Abstract: The controller is one of the most important modules in the autonomous driving pipeline, ensuring the vehicle reaches its desired position. In this work, a reinforcement learning based lateral control approach, despite the imperfections in the vehicle models due to measurement errors and simplifications, is presented. Our approach ensures comfortable, efficient, and robust control performance considering the interface between controlling and other modules. The controller consists of the conventional Model Predictive Control (MPC)-PID part as the basis and the demonstrator, and the Deep Reinforcement Learning (DRL) part which leverages the online information from the MPC-PID part. The controller’s performance is evaluated in CARLA using the ground truth of the waypoints as inputs. Experimental results demonstrate the effectiveness of the controller when vehicle information is incomplete, and the training of DRL can be stabilized with the demonstration part. These findings highlight the potential to reduce development and integration efforts for autonomous driving pipelines in the future.

Keywords: Reinforcement Learning for Planning, Feedback Systems for Driver Interaction, Predictive Trajectory Models and Motion Forecasting
Abstract: Integrating human expertise with machine learning is crucial for applications demanding high accuracy and safety, such as autonomous driving. This study introduces Interactive Double Deep Q-network (iDDQN), a Human-in-the-Loop (HITL) approach that enhances Reinforcement Learning (RL) by merging human insights directly into the RL training process, improving model performance. Our proposed iDDQN method modifies the Q-value update equation to integrate human and agent actions, establishing a collaborative approach for policy development. Additionally, we present an offline evaluative framework that simulates the agent’s trajectory as if no human intervention to assess the effectiveness of human interventions. Empirical results in simulated autonomous driving scenarios demonstrate that iDDQN outperforms established approaches, including Behavioral Cloning (BC), HG-DAgger, Deep Q-Learning from Demonstrations (DQfD), and vanilla DRL in leveraging human expertise for improving performance and adaptability.

Keywords: End-to-End Neural Network Architectures and Techniques, Reinforcement Learning for Planning, Motion Forecasting
Abstract: This paper addresses the challenges of training end-to-end autonomous driving agents using Reinforcement Learning (RL). RL agents are typically trained in a fixed set of scenarios and nominal behavior of surrounding road users in simulations, limiting their generalization and real-life deployment. While Domain Randomization offers a potential solution by randomly sampling driving scenarios, it frequently results in inefficient training and sub-optimal policies due to the high variance among training scenarios. To address these limitations, we propose an automatic curriculum learning framework that dynamically generates driving scenarios with adaptive complexity based on the agent's evolving capabilities. Unlike manually designed curricula that introduce expert bias and lack scalability, our framework incorporates a "teacher" that automatically generates and mutates driving scenarios based on their learning potential—an agent-centric metric derived from the agent's current policy, eliminating the need for expert design. The framework enhances training efficiency by excluding scenarios the agent has mastered or finds too challenging. We evaluate our framework in a reinforcement learning setting where the agent learns a driving policy from camera images. Comparative results against baseline methods, including fixed scenario training and domain randomization, demonstrate that our approach leads to superior generalization, achieving higher success rates (+9% in low traffic density, +21% in high traffic density) and faster convergence with fewer training steps. Our findings highlight the potential of ACL in improving the robustness and efficiency of RL-based autonomous driving agents.

Keywords: Adaptive Vehicle Control Techniques, Reinforcement Learning for Planning, Level 2 ADAS Control Techniques
Abstract: Although significant progress has been made in autonomous vehicle control, particularly in maneuvering beyond traction limits using methods like reinforcement learning, the critical challenge of applying these closed-loop control techniques to real-world conditions persists and necessitates further research. To advance this field, the primary objective of this paper is to explore an approach for reinforcement learning based self-driving agents to perform circular drift maneuvers under rapidly and uncertainly changing environmental conditions. The agents were trained in simulation using robust adversarial reinforcement learning (RARL) to improve robustness against significant disturbances in the dynamics of the drivetrain. Agents trained with RARL turned out to be superior to those trained without this technique, providing less uncertainty when exposed to such disturbances.

Keywords: Real-Time Control Strategies, Decision Making, Reinforcement Learning for Planning
Abstract: Lane-free traffic is a new concept proposed for the era of connected and automated vehicles (CAVs). In this system, vehicles are no longer restricted to traditional lanes, and any lateral location within the entire road boundaries is considered for navigation. In the current lane-based traffic system, roundabouts, characterized by wide lanes or no clear lane markings, allow vehicles to have more lateral movement freedom and thus offer an ideal setting to investigate how CAVs should behave in lane-free conditions. This study introduces a new approach to controlling CAVs in a lane-free urban environment using Deep Reinforcement Learning (DRL). This constitutes the first time DRL has been applied to help intelligent vehicles drive through urban roundabouts without the constraints of traditional lanes. By allowing vehicles to use the entire road space, the model aims to provide a comfortable and collision-free driving experience, enabling vehicles to maintain desired speeds. Our methodology involves developing a Deep Deterministic Policy Gradient (DDPG) based control strategy that enables CAVs to make dynamic, real-time decisions for efficient navigation, merging, and exiting. To test this approach, we simulated a real-world roundabout, already deploying a lane-free design. We applied our model to all CAVs driving under various traffic patterns in that environment. We also compared its performance to a two-dimensional control strategy based on the self-driven particle model for mixed traffic. The findings indicate that our DRL-employing autonomous vehicles are able to learn smooth driving policies and achieve target speeds, in addition to avoiding collisions and ensuring a comfortable experience for passengers.

Keywords: Reinforcement Learning for Planning, Collision Avoidance Algorithms, Decision Making
Abstract: Reinforcement learning (RL) has become a key method for decision-making in autonomous vehicles, particularly in tasks like path planning and obstacle avoidance. However, RL often struggles with sparse or delayed reward signals, which can hinder learning. A promising solution is incorporating a teacher-student framework, where the agent learns from a teacher. While these methods can accelerate learning, they risk sub-optimal outcomes if the teacher’s guidance is flawed. This paper introduces Double Entropy Reinforcement Learning (DERL), which enables the student to initially rely on the teacher and gradually shift towards independent exploration as it surpasses the teacher’s performance. Experiments show that DERL outperforms existing teacher-student algorithms in learning efficiency and effectiveness, reducing reliance on imperfect guidance.

Keywords: Collision Avoidance Algorithms, Predictive Trajectory Models and Motion Forecasting, Advanced Multisensory Data Fusion Algorithms
Abstract: This paper presents a novel autonomous navigation framework that integrates Large Language Models (LLMs) with multimodal sensor fusion to enable dynamic obstacle avoidance and human-aware path planning in diverse environments. The proposed system leverages an FPGA-accelerated fusion pipeline, combining LiDAR and vision data for real-time perception. A Hungarian algorithm-based object matching technique ensures robust tracking, while a bird’s-eye view (BEV) representation enhances spatial reasoning and occlusion handling. The fused sensory inputs are processed by a fine-tuned LLM, which contextualizes pedestrian behavior and environmental constraints to generate adaptive, human-centric navigation strategies. Unlike traditional rule-based methods, LLMs provide generalization capabilities to novel scenarios, significantly improving interaction with vulnerable pedestrians such as children, elderly individuals, and wheelchair users. Extensive evaluations in both simulated and real-world scenarios confirm the system’s ability to reduce collisions and enhance navigation efficiency in high-density environments. By bridging semantic reasoning and robotic control, this work lays the foundation for next-generation intelligent navigation systems that are both safety-aware and scalable across autonomous platforms.

Keywords: Motion Planning Algorithms for Autonomous Vehicles, Decision Making
Abstract: Driving autonomously in diverse environments remains a significant challenge, especially when transitioning between regions with distinct traffic cultures and regulations. Whilst human drivers exhibit remarkable adaptability through experiential learning and cognitive modeling, current data-driven autonomous systems often struggle with environmental adaptation, interpretability, and continuous learning capabilities. In this work, we present LoDriver (Local Driver), a novel knowledge-based architecture that enhances the conventional scene understanding-decision-planning paradigm through cognitive-inspired dual-process reasoning for path planning. LoDriver integrates a reactive process with a deliberative mechanism, processing multi-modal scene descriptions through parallel pathways. It maintains a structured memory module that dynamically accumulates driving experiences, traffic regulations, and knowledge, enabling experience-based decision-making and continuous learning through systematic memory updates. Experimental evaluation on the nuScenes dataset demonstrates LoDriver's interpretability and enhanced performance compared to existing knowledge-driven models, highlighting its advantage in operating across different environments.

Keywords: Vehicle-to-Infrastructure (V2I) Communication
Abstract: Road Side Units (RSUs) play a pivotal role in enhancing the safety of Connected Vehicles (CVs), yet their safety benefits hinge significantly on effective deployment strategies. Traditional approaches often focus on high-risk areas, but such locations may not necessarily yield the most substantial safety improvements. This study redefines hot spots as road segments where RSU deployment results in the greatest reduction of collision risk and introduces a cost-efficient method for identifying such locations. The proposed method requires only small-scale real-world driving data, which is further augmented to support broader scenario evaluation. Additionally, an accelerated sampling strategy is incorporated to enhance the efficiency of the identification process. Simulation-based evaluations demonstrate that the method achieves superior performance in terms of safety impact, data efficiency, compared to conventional approaches.

Keywords: Level 4-5 Autonomous Driving Systems Architecture, Decision Making, Real-World Testing Methodologies for Safety Systems
Abstract: Autonomous vehicles (AVs) promise significant advances in transportation safety and efficiency. However, navigating roadway work zones, which can be rather complex and dynamic, remains a significant challenge. This paper presents the results of a large study that addresses the challenges, requirements, solutions and practical experiences of AVs driving safely through work zones. We begin by proposing a taxonomy of work zone scenarios, analyzing their types and attributes. We next discuss the perception, routing, behavioral and path-planning requirements for AVs to safely navigate these scenarios. We then offer methods to meet these requirements and investigate the impact of range and AV speed on perception confidence levels for work zone detection. We evaluate our solutions in a co-simulation environment, on a closed test-track and on public roadways across more than 20 work zone scenarios specified by the Pennsylvania Department of Transportation (PennDoT). Video demonstrations illustrate the feasibility of safe and reliable navigation of AVs in a wide variety of work zones.

Keywords: Level 4-5 Autonomous Driving Systems Architecture
Abstract: This paper presents the design and validation of our autonomous driving system developed for one-on-one high-speed racing at the Indy Autonomous Challenge (IAC) 2024. In this event, each team is required to pass the opponent car on an oval racetrack at very high speeds reaching more than 250 km/h. In order to meet the critical challenges of high-speed autonomous racing, we developed reliable perception, behavior planning for passing, and control algorithms stable in the high speed range. For perception, we constructed a multi-modal pipeline using cameras, LiDARs, and radars to achieve accurate and reliable real-time object detection. We also designed a new state machine-based algorithm for effective planning of overtaking maneuvers. As for the control, we improved the vehicle controller at higher speed ranges by building a more accurate engine torque map and fine-tuning the controller during the actual runs. Our newly designed racing system was proven effective, enabling our car to cruise stably at 250.1 km/h in solo lap and executing multiple successful overtaking maneuvers during the one-on-one races at the Las Vegas Motor Speedway (LVMS) for the Consumer Electronics Show (CES) 2024.

Keywords: Electric and Hybrid Vehicle Integration, Decision Making, Automotive Datasets
Abstract: Accurately estimating the driving range of electric trucks remains a critical challenge due to higher complexity and operational diversity compared to passenger EVs. Factors such as variable payloads, aerodynamic variability from different trailers, and auxiliary power demands significantly impact energy usage. Conventional data-driven models, while effective, frequently struggle to generalize under extreme conditions. This paper introduces a novel Physics-Informed Neural Network (PINN) framework for electric truck range estimation that combines physics-based modeling with data-driven learning. A custom loss function dynamically balances physical constraints and empirical insights using a time-varying parameter. The model prioritizes physics early in training, then shifts toward datadriven fine-tuning. Experimental results demonstrate that while pure data-driven models achieve higher accuracy under normal conditions, the proposed PINN framework demonstrates superior robustness, particularly with small datasets and in challenging weather and operational scenarios. These findings highlight the potential of PINN-based methods in addressing key challenges in electric truck range estimation.

Keywords: Smart City Mobility Integration Strategies, Cooperative Planning Strategies in Vehicle Networks
Abstract: This paper presents a digital twin-empowered real-time optimal delivery system specifically validated through a proof-of-concept (PoC) demonstration of a real-world autonomous car-sharing service. This study integrates real-time data from roadside units (RSUs) and connected and autonomous vehicles (CAVs) within a digital twin of a campus environment to address the dynamic challenges of urban traffic. The proposed system leverages the Age of Information (AoI) metric to optimize vehicle routing by maintaining data freshness and dynamically adapting to real-time traffic conditions. Experimental results from the PoC demonstrate a 22% improvement in delivery efficiency compared to conventional shortest-path methods that do not consider information freshness. Furthermore, digital twin-based simulation results demonstrate that this proposed system improves overall delivery efficiency by 12% and effectively reduces the peak average AoI by 23% compared to the conventional method, where each vehicle selects the shortest route without considering information freshness. This study confirms the practical feasibility of cooperative driving systems, highlighting their potential to enhance smart mobility solutions through scalable digital twin deployments in complex urban environments.

Keywords: Deep Learning Based Approaches, Electric and Hybrid Vehicle Integration
Abstract: Battery health estimation is fundamental to ensure battery safety and reduce cost. However, achieving accurate estimation has been challenging due to the batteries' complex nonlinear aging patterns and capacity regeneration phenomena. In this paper, we propose OSL, an optimal signal decomposition-based multi-stage machine learning for battery health estimation. OSL treats battery signals optimally. It uses optimized variational mode decomposition to extract decomposed signals capturing different frequency bands of the original battery signals. It also incorporates a multi-stage learning process to analyze both spatial and temporal battery features effectively. An experimental study is conducted with a public battery aging dataset. OSL demonstrates exceptional performance with a mean error of just 0.26%. It significantly outperforms comparison algorithms, both those without and those with suboptimal signal decomposition and analysis. OSL considers practical battery challenges and can be integrated into real-world battery management systems, offering a good impact on battery monitoring and optimization.

Keywords: Trust and Acceptance of Autonomous Technologies, Level 4-5 Autonomous Driving Systems Architecture, Smart City Mobility Integration Strategies
Abstract: This paper introduces SumoWare, an interface integrating state-of-the-art platforms, SUMO and Autoware, to enable realistic evaluation of autonomous vehicle (AV) behavior in diverse traffic scenarios. Addressing the critical need for flexible and reproducible testing environments, SumoWare establishes seamless bidirectional communication, ensuring precise synchronization between the two systems. Experimental results demonstrate strong spatiotemporal consistency between the platforms and, based on a case study, reveal the implications of default conservative AV behavior on traffic flow, including reduced capacity and earlier congestion. These findings highlight the pressing need to refine planning and control strategies to mitigate the overly cautious nature of AV systems and optimize their integration into real-world traffic.

Keywords: Infrastructure Requirements for Automated Vehicles, Smart City Mobility Integration Strategies, Cooperative Planning Strategies in Vehicle Networks
Abstract: Green Wave provides practical and advanced solutions to improve traffic efficiency and safety through network coordination. Nevertheless, the complete potential of Green Wave systems has yet to be explored. Using emerging technologies and advanced algorithms like AI or V2X would aid in achieving more robust traffic management strategies, especially when integrated with Green Wave. This work comprehensively surveys existing traffic control strategies enabling Green Waves and analyses their impact on future traffic management systems and urban infrastructure. Understanding previous research on traffic management and its effect on traffic efficiency and safety helps explore the integration of Green Wave solutions with smart city initiatives for effective traffic signal coordination. This paper also discusses the advantages of using Green Wave strategies for emission reduction and considers road safety issues for vulnerable users like pedestrians and cyclists. Finally, the existing challenges and research gaps in building robust and successful Green Wave systems are discussed to articulate explicitly the future requirement of sustainable urban transport.

Keywords: Level 2 ADAS Control Techniques, User-Centric Intelligent Vehicle Technologies, User Experience in Autonomous Vehicles
Abstract: With the rapidly advancing capabilities of Level 2 automated systems the need for driver confirmation of complex maneuvers, such as entering roundabouts, has been increasingly emphasized to ensure active engagement and safety. This study contrasts confirming maneuvers through control gazes with traditional explicit methods like accelerator use to facilitate seamless interaction and promote the adoption of L2 systems. A simulator study involving 64 participants demonstrated a clear preference for the gaze confirmation system, with significantly higher ratings for user experience, acceptance, and intention to use, without compromising trust compared to explicit confirmation. Participants intervened in case of system failures with either system. The results are building a strong case for daring advanced automation via gaze confirmation to harness progress of L2 systems while preserving safety.

Keywords: Infrastructure Requirements for Automated Vehicles, Smart City Mobility Integration Strategies
Abstract: Advancements in Autonomous Vehicles (AVs) technology necessitate Active Traffic Management (ATM) strategies that incorporate fine-grained microscopic traffic features, accounting for AVs’ distinct driving behaviors, decision-making patterns, and the specific road geometry design required for increasing AV penetration. While microscopic traffic simulation models are well-suited for capturing these microscopic features, their high computational demands and non-analytic nature have confined their use to ATM strategies optimization. To bridge this gap, this study employs a Simulation-Based Optimization (SBO) paradigm to integrate microscopic models into the design of ATM strategies for highways. We introduce an SBO framework that combines a Cell Transmission Model (CTM)-based meta model with the Adaptive Hyperbox Algorithm (AHA), where the CTM-based meta model approximates microscopic traffic conditions, and the AHA explores high-dimensional, discrete decision spaces to determine optimal ATM strategies.

Our experiments demonstrate that the proposed SBO framework substantially outperforms other general-purpose SBO methods. This work advances ATM optimization by facilitating the integration of microscopic simulation models into practical optimization frameworks, ensuring the consideration of evolving traffic characteristics with the increasing presence of AVs.

Keywords: Safety Verification and Validation Techniques, Level 4-5 Autonomous Driving Systems Architecture, Level 3 Driving Systems Architecture and Techniques
Abstract: Developers of an Automated Driving System (ADS) must demonstrate that the system can operate safely within its designated operational environment. A key aspect of this is the Operational Design Domain (ODD), which defines the conditions under which an ADS is designed to operate safely. The ODD utilizes concepts from a taxonomy that formally represent environmental conditions. However, not all concepts within a taxonomy are relevant for the development of an ADS. Current approaches defining an ODD often neglect the systematic identification of relevant and irrelevant concepts. This omission poses a significant challenge to comprehensive system development, as ensuring the inclusion of only relevant concepts in the ODD contributes to a more efficient and streamlined development process. Addressing this requires explicitly, evaluating both relevant and irrelevant concepts, accompanied by justification ensures transparency and traceability. Hence, we present a novel framework that supports the systematic identification of relevant taxonomy concepts for creating an ODD. Additionally, this framework serves as a critical component in establishing a safety-by-design development process by incorporating structured documentation to support the argumentation for including or omitting specific taxonomy concepts. Through a case study, we demonstrate that the framework provides comprehensive and robust justification, streamlines relevance argumentation across taxonomy concepts, and contributes to a reduced ODD validation effort.

Keywords: Level 4-5 Autonomous Driving Systems Architecture, Level 3 Driving Systems Architecture and Techniques
Abstract: Automated driving systems (ADS) have significant potential to make mobility safer and more efficient as they evolve and integrate new technologies. These advances continuously change the operational boundaries where ADS can operate safely as specified by the operational design domain (ODD). Nevertheless, research into ADS and ODD is still in its infancy, so that long-term concepts that cover the entire product life cycle are not yet available. Current research focuses on the initial development of ADS and ODD. A more comprehensive perspective, particularly regarding the long-term evolution of an ODD during operation and throughout the life cycle of an ADS, remains largely unaddressed. This paper describes an overview of the life cycle of an ADS and its corresponding ODD that consists of initial development, ongoing versioning, and end-of-life management. Furthermore, the paper examines how the definition and implementation of an ODD, as an essential asset for ADS development, influence the phases identified in the ADS life cycle to ensure alignment with the requirements of the stakeholders.

Keywords: Automotive Datasets, Human Factors Analysis in Vehicle Design, Level 2 ADAS Control Techniques
Abstract: Understanding and modeling human driving behavior is essential to improve advanced driver assistance systems (ADAS) and traffic simulations. Traditional driver models are often limited in capturing context- and environment-dependent variations due to simplifying assumptions. The increasing availability of real-world driving data from connected vehicles enables the identification of context-adaptive model parameters, provided that strict privacy and anonymization measures are observed. We analyze data from 156 electric vehicles, each equipped with high-frequency data loggers, recording over 2000 signals per vehicle during a six-month period in uncontrolled environments. As a case study, we examine the approach behavior of a preceding vehicle, which is important for adaptive cruise control (ACC), and its dependence on temperature, light levels, and rain. Our findings suggest that cold temperatures and nighttime conditions increase time-headway and following distances, while the effects of rain appear less pronounced. In the long term, such datasets could refine context-aware driver models for simulation and ADAS calibration and enable personalized adaptive cruise control based on learned driving preferences. By integrating context- and environment-aware driver models, this approach bridges the gap between theoretical modeling and real-world driving behavior.

Keywords: Feedback Systems for Driver Interaction, User-Centric Intelligent Vehicle Technologies
Abstract: Steer-by-Wire is an innovative key enabler for autonomous driving in intelligent future vehicle platforms. By controlling steering systems electronically without a mechanical link between steering wheel and front axle, one main challenge is the dynamic reproduction of road excitation as steering torque feedback for the driver. In this work, two excitation scenarios are investigated using measured data from a testing vehicle, as well as simulated data from a multibody simulation. The resulting simulations and measurements are compared, and driving dynamic signals are examined for correlation with steering torque and rack force. The chronological sequence of these signals is analyzed, allowing the identification of lateral and vertical driving dynamic signals as promising bases for steering torque feedback generation.

Keywords: Level 4-5 Autonomous Driving Systems Architecture
Abstract: Abstract—This paper investigates Reinforcement Learning (RL) for model-based static output feedback, with a focus on vehicle lateral control applications such as lane-keeping and lane-change maneuvers. Emphasizing stability during learning to prevent unsafe actions, the study reviews recent RL advancements and compares policy iteration (PI) and policy search (PS) methods in terms of convergence speed and stability. A two-degree-of-freedom bicycle model is employed to validate the proposed PI algorithm for static output feedback, highlighting its faster convergence compared to PS and its competitive performance against traditional Linear Quadratic Regulator (LQR) control. Results demonstrate the practicality and efficiency of RL for stable real-time vehicle control

Keywords: Motion Planning Algorithms for Autonomous Vehicles, Smart City Mobility Integration Strategies, Adaptive Vehicle Control Techniques
Abstract: Autonomous Freight Trucks enhance freight delivery efficiency in open logistics areas but face challenges when coexisting with human-driven trucks. This study proposes a spatiotemporal right-of-way allocation approach, enabling AFTs with different turning movements to alternately share a Truck Priority Lane. Experiments show the approach improves delivery frequency by 27.7%–31.6%, and freight throughput by 10.3%–15.1%. Benefits grow with higher AFT penetration rates, demonstrating robust performance under mixed traffic conditions.

Keywords: Level 4-5 Autonomous Driving Systems Architecture, Motion Planning Algorithms for Autonomous Vehicles, Teleoperation Control Systems for Vehicles
Abstract: We present a vehicle system that is able to navigate safely and efficiently around Vulnerable Road Users (VRUs), such as pedestrians and cyclists. We describe the various system modules (environment perception, localization and mapping, motion planning, and control), the system architecture, and our prototype vehicle. One distinguishing aspect is a novel motion planner, based on Topology-driven Model Predictive Control (TMPC). The guidance planner computes several trajectories in parallel that each pass obstacles in distinct way, or are nonpassing. The underlying trajectory optimization constrains the joint probability of collision with VRUs under generic uncertainties. To account for extraordinary situations (“edge cases”) that go beyond the autonomous capabilities — such as construction zones or encounters with emergency responders — the system includes an option for remote human operation, supported by visual and haptic guidance.

In experiments in simulation, we show that our motion planner outperforms three baselines in terms of safety and efficiency. Furthermore, we describe tests with the prototype vehicle on a test track in self-driving mode and with remote operation.

Keywords: Trust and Acceptance of Autonomous Technologies, Ethics in Driving, Data Sharing and Privacy in V2X Systems
Abstract: As autonomous vehicle (AV) technology advances, ethical questions surrounding its implementation and integration become increasingly apparent. This work explores the potential of leveraging CARLA, an open-source driving simulator for AVs in urban environments, to simulate ethical dilemmas that could occur in real-world self-driving car scenarios. A scenario depicting a more likely, less emotionally loaded ethical dilemma was desired in response to AV ethics' seeming overfocus on the Trolley Problem. Additionally, the framing of the chosen dilemma as a trade-off became a central concept and contribution of the work. Brief reviews of the existing AV ethics literature and key technologies are provided, and similar works described. Then, a scenario involving an ambulance exiting a congested highway is defined and implemented with the specific ethical trade-off of privacy vs. efficiency in mind. Vehicle-to-vehicle (V2V) communication levels are used as markers of privacy, while the elapsed simulation time for the ambulance to reach its goal measures efficiency. Thus, the privacy-efficiency trade-off is evaluated at different levels of privacy sacrifice, where privacy can be discussed qualitatively and efficiency quantitatively. Experimental results determine that the simulation grows more efficient as privacy sacrifice increases, though partly due to biases in devised agent behaviors. The use of a trade-off for analysis in combination with simulation software is emphasized as the most significant contribution and encouraged for future AV ethics work.

Keywords: Electric and Hybrid Vehicle Integration, Smart City Mobility Integration Strategies, Vehicle-to-Infrastructure (V2I) Communication
Abstract: Eco-approach and Departure (EAD) strategies optimize energy efficiency in Connected and Automated Vehicles (CAVs) by leveraging Signal Phase and Timing (SPaT) data to adjust trajectories. This paper extends the EAD strategy to incorporate actuated signals in a city-scale electrified network and introduces the Eco-Pedaling (EP) strategy to further reduce electricity consumption. First, we analyze off-peak and peak scenarios, applying eco-driving technologies to Connected and Automated Electric Vehicles (CAEVs) in the city network, assuming all vehicles are electric. Results show that CAEVs with eco-driving technologies achieve a 32.24% reduction in electricity consumption per mile during off-peak hours and 31.50% during peak hours, with a 15% CAEV penetration rate, corresponding to Toyota’s market share in North America. Second, a 24-hour evaluation with the same CAEV penetration rate highlights a 6.02% reduction in total electricity consumption across the network, at the cost of a 3.58% decrease in network efficiency. Finally, analysis of varying CAEV penetration rates reveals that the system-level electricity consumption decreases as the proportion of CAEVs increases.

Keywords: Motion Planning Algorithms for Autonomous Vehicles
Abstract: The classical g-g diagram, representing the achievable acceleration space for a vehicle, is commonly used as a constraint in trajectory planning and control due to its computational simplicity. To address non-planar road geometries, this concept can be extended to incorporate g-g constraints as a function of vehicle speed and vertical acceleration, commonly referred to as g-g-g-v diagrams. However, the estimation of g-g-g-v diagrams is an open problem. Existing simulation-based approaches struggle to isolate non-transient, open-loop stable states across all combinations of speed and acceleration, while optimization-based methods often require simplified vehicle equations and have potential convergence issues. In this paper, we present a novel, open-source, quasi-steady-state black box simulation approach that applies a virtual inertial force in the longitudinal direction. The method emulates the load conditions associated with a specified longitudinal acceleration while maintaining constant vehicle speed, enabling open-loop steering ramps in a purely QSS manner. Appropriate regulation of the ramp steer rate inherently mitigates transient vehicle dynamics when determining the maximum feasible lateral acceleration. Moreover, treating the vehicle model as a black box eliminates model mismatch issues, allowing the use of high-fidelity or proprietary vehicle dynamics models typically unsuited for optimization approaches. An open-source version of the proposed method is available at: https://github.com/TUM-AVS/GGGVDiagrams

Keywords: Teleoperation Control Systems for Vehicles, User Experience in Autonomous Vehicles, Real-World Testing Methodologies for Safety Systems
Abstract: Mission-critical services such as remote vehicle control necessitate real-time interactions between vehicles and cloud services that can be realized using next-generation mobile networks. A previous study proposed a Real-Time Simulator (RTS) coordinated by using three simulators for mission-critical service: CARLA, OMNeT++, and SUMO, and evaluated the missioncritical services within a single mobile network. Carrier-bonding communication uses multiple mobile networks to achieve highly reliable communication. Thus, it is a promising candidate for enhancing service reliability and should be evaluated using such a RTS. However, the deployment of multiple mobile networks in an OMNeT++ instance is difficult owing to the significant increase in the computational burden of processing large amounts of communication traffic. Thus, this study propose an advanced RTS architecture for the evaluation of highly reliable communication. In the RTS architecture, multiple OMNeT++ instances are employed to evaluate the carrier-bonding communication. Referring to the micro-service architecture, the proposed RTS employs Message Queueing Telemetry Transport to share vehicle position information asynchronously among all simulators. Further, LXD container virtualization is used to ensure the availability of the processing resources for each OMNeT++. The proposed RTS ensures scalability while maintaining the throughput performance for each OMNeT++ instance. Furthermore, the RTS is capable of evaluating mission-critical services through carrier-bonding communication.

Keywords: Safety Verification and Validation Techniques, Vulnerable Road User Protection Strategies, Real-Time Control Strategies
Abstract: While several commercial software packages for simulating the dynamics of robot or human-ridden Single-Track Vehicles (STVs) such as bicycles, motorcycles, or other Powered Two-Wheelers (PTWs) are common in the literature, open-source options for high-fidelity simulations of STV dynamics are limited. This paper explores whether the open-source Webots robot simulator is capable of representing the dynamics of STVs with sufficient fidelity to allow for robotic rider design. Data from a small-scale self-balancing PTW are compared with both a linear dynamic model and with simulation results from a nonlinear, multi-body model of the vehicle custom-built in Webots. Results indicate that Webots's physics engine provides sufficient dynamic fidelity to accurately represent the behavior of the vehicle in both simple step response tests and in assisted teleoperation with automatic starting and stopping via a motorized kickstand.

Keywords: Real-World Testing Methodologies for Safety Systems, Vulnerable Road User Protection Strategies, Predictive Trajectory Models and Motion Forecasting
Abstract: The interaction performance of autonomous vehicles (AVs) is of significant importance, especially in scenarios where AVs interact with two-wheelers at intersections. These scenarios are characterized by high exposure and hazard rates, and the behavior of two-wheelers in such scenarios is complex and variable. Therefore, it is necessary to accurately model the behavior of two-wheelers in severe conflict scenarios to support more reasonable and comprehensive testing results. In this paper, a two-wheeler driving behavior model (Improved Social Force Model for Simulation, ISFM4Sim) is proposed for simulation testing of AVs. On one hand, in order to apply the model to intersections with similar topology, the boundary repulsive forces are removed. The error in the ablation experiment result only increased by about 0.27%, which suggests that the forces have little contribution in severe conflict scenarios. On the other hand, by adding a dynamics module and modifying the model output to control signals, it is realized to apply the model in a simulation environment. Finally, simulation experiments demonstrate that ISFM4Sim performs well in bidirectional interactions and has good potential for testing of AVs.

Keywords: Safety Verification and Validation Techniques, Real-World Testing Methodologies for Safety Systems
Abstract: Autonomous driving systems (ADS) shall safely navigate diverse scenarios encountered in an open world. Prior to deployment, it is crucial to evaluate their ability to handle these situations. A popular approach for ADS verification is scenario-based testing, which simulates relevant critical scenarios in a virtual environment. However, evaluating these behaviors can be challenging, as traditional scenario-based testing often relies on metrics that may overlook subtle risks without immediate critical impact that still pose long-term safety concerns. System-theoretic process analysis (STPA) is a hazard analysis technique suitable for the analysis of modern complex systems. In this work, we employ STPA to derive system-level evaluation criteria for scenario-based testing frameworks designed to be universally applicable in any simulation run. Their utilization provides a data-driven approach to safety assurance, helping to uncover unknown safety issues that, if unaddressed, could compromise the long-term reliability of ADS. Furthermore, the approach is demonstrated on a construction zone navigation extension for ADS, illustrating its potential to provide continuous, data-driven feedback from testing to safety assurance. The method shall contribute to the early detection of arising safety issues and ultimately lead to the development of safer systems.

Keywords: Teleoperation Control Systems for Vehicles, User-Centric Intelligent Vehicle Technologies, Infrastructure Requirements for Automated Vehicles
Abstract: Implementing a teleoperation system with its various actors and interactions is challenging and requires an overview of the necessary functions. This work collects all tasks that arise in a control center for an automated vehicle fleet from literature and assigns them to the two roles Remote Operator and Fleet Manager. Focusing on the driving-related tasks of the remote operator, a process is derived that contains the sequence of tasks, associated vehicle states, and transitions between the states. The resulting state diagram shows all remote operator actions available to effectively resolve automated vehicle disengagements. Thus, the state diagram can be applied to existing legislation or modified based on prohibitions of specific interactions. The developed control center framework and included state diagram should serve as a basis for implementing and testing remote support for automated vehicles to be validated on public roads.

Keywords: Safety Verification and Validation Techniques, Multi-Agent Coordination Strategies, Level 3 Driving Systems Architecture and Techniques
Abstract: Virtual simulation is becoming an essential part of the development of autonomous driving vehicles. Even with the recent advancements for virtual simulation frameworks, less efforts have been made to enable timing-accurate multi-vehicle simulations. In this regard, we present a virtual simulation environment for truck platooning scenarios, where the simulation engine has to calculate the dynamics and environment change of multiple trucks. In such multi-vehicle simulations, the computation for simulation easily exceeds the given simulation hardware's computing capacity, making real-time simulations difficult to achieve. As a solution, we provide a synchronous multi-agent simulation framework that can synchronize the simulation engine's progress with the algorithm computations of each truck. Our synchronous simulation framework guarantees deterministic simulation results, however, at the cost of increased simulation running time. To alleviate the synchronization overhead, we also propose a shared memory-based massive data interface. Our method is implemented and evaluated based on the CARLA autonomous driving simulator.

Keywords: Techniques for Dataset Domain Adaptation, Synthetic Data Generation for Training
Abstract: Simulating realistic LiDAR intensity is essential for autonomous driving, particularly under snow conditions, where current methods fail to capture complex LiDAR-toatmosphere interactions. This paper introduces a CycleGAN framework guided by physics, which incorporates the principles of LiDAR intensity attenuation in snowy weather, significantly narrowing the simulation-to-reality gap. The model was evaluated using an open-source real snow dataset and an opensource simulated dataset, demonstrating its ability to replicate real-world intensity patterns with high accuracy, as indicated by metrics like Structural Similarity Index Measure (SSIM), Kullback-Leibler (KL) Divergence, etc. In the downstream semantic segmentation task, models trained on the enhanced data outperformed those trained on baseline datasets, underscoring the framework’s effectiveness in improving LiDAR data realism and robustness in snow-weather autonomous driving scenarios

Keywords: Synthetic Data Generation for Training, Integration Methods for HD Maps and Onboard Sensors, Level 4-5 Autonomous Driving Systems Architecture
Abstract: Autonomous racing has emerged as a crucial testbed for autonomous driving algorithms, necessitating a simulation environment for both vehicle dynamics and sensor behavior. Striking the right balance between vehicle dynamics and sensor accuracy is crucial for pushing vehicles to their performance limits. However, autonomous racing developers often face a trade-off between accurate vehicle dynamics and high-fidelity sensor simulations. This paper introduces R-CARLA, an enhancement of the CARLA simulator that supports holistic full-stack testing, from perception to control, using a single system. By seamlessly integrating accurate vehicle dynamics with sensor simulations, opponents simulation as Non-Player-Characters (NPCs), and a pipeline for creating digital twins from real-world robotic data, R-CARLA empowers researchers to push the boundaries of autonomous racing development. Furthermore, it is developed using CARLA’s rich suite of sensor simulations. Our results indicate that incorporating the proposed digital-twin framework into R-CARLA enables more realistic full-stack testing, demonstrating a significant reduction in the Sim-to-Real gap of car dynamics simulation by 42% and by 82% in the case of sensor simulation across various testing scenarios.

Keywords: Human Factors Analysis in Vehicle Design, Synthetic Data Generation for Training, Vulnerable Road User Protection Strategies
Abstract: Understanding pedestrian behavior is crucial for improving road safety especially considering the advent of autonomous vehicles (AVs) and AD/ADAS systems. This study employs a Virtual Reality (VR) simulator to investigate pedestrian crossing decisions under varying vehicle speeds (20, 30, and 40 km/h) and Time-to-Arrival (TTA) intervals (3, 4, 5, and 6 seconds). Key behavioral metrics, including crossing percentage, gap acceptance, safety margin, and time to crossing decision, were analyzed.

The results indicate that longer TTAs significantly increase crossing likelihood, as pedestrians perceive greater safety. Although higher vehicle speeds also correlate with increased crossing percentages, this effect may stem from the experimental design, where higher speeds resulted in greater physical distances to the vehicle. Statistical analysis confirmed that TTA strongly influences pedestrian safety metrics, particularly safety margin and time to crossing decision, while vehicle speed had a limited effect.

Qualitative feedback highlighted challenges in speed estimation, particularly at greater distances, which likely influenced crossing decisions. Participants also reported minor limitations related to VR immersion, such as the absence of auditory cues and concerns about the headset cable. These factors may have influenced crossing behavior, underscoring the importance of refining VR environments for pedestrian studies.

Keywords: Behavior Assessment Using Cooperative Data, Feedback Systems for Driver Interaction, Human-Machine Interface (HMI) Design Principles
Abstract: Interaction modeling plays a huge role in understanding human behavior in traffic. This is especially relevant when it comes to interactions between vehicles and vulnerable road users such as pedestrians. Thus, pedestrian intention prediction is an ongoing field of research in order to understand the pedestrians' decision making. Most state-of-the-art prediction frameworks are trained on large-scale datasets and evaluated with respect to acknowledged benchmarks. These datasets lack the ability to account for the reciprocal nature of interactions between pedestrians and vehicles and the effects of the two agents influencing each other. In this work, we demonstrate first steps towards assessing pedestrian prediction algorithms within realistic scenarios including the interaction effects arising from its interplay with a planning component. For this, we validate an existing prediction framework trained on benchmark datasets with situations from a virtual reality (VR) pedestrian-vehicle co-simulator that allows us to include the effect of vehicle planning on pedestrian behavior. We evaluate the performance of the prediction framework comparing data from pre-recorded real-world datasets with data from our co-simulation study and conduct an ablation analysis to identify the most important features for pedestrian intention prediction. The results highlight the significance of pedestrian action and proximity to the road.

Keywords: Synthetic Data Generation for Training, 3D Scene Reconstruction Methods, Safety Verification and Validation Techniques
Abstract: The development of autonomous vehicles (AV) and advanced driver assistance systems (ADAS) relies on effective perception systems to interpret complex environmental scenarios through a robust combination of sensors. Traditional road testing is limited in fully exploring the operational design domain, especially in rare or hazardous conditions. This study investigates the viability of sensor simulation as a complementary evaluation method, focusing on lidar technology. A "digital twin" is an accurate simulation model replicating real-world sensor interactions and behaviors. Through detailed analysis of synthetic lidar simulations and real-world drive logs, we find that digital twins can effectively replicate real-world scenarios, albeit with some variance. This paper discusses the technical challenges and limitations in achieving high-fidelity simulations and the potential of these technologies to improve the safety, reliability, and efficiency of autonomous systems. Unlike previous approaches, our pipeline provides an end-to-end solution from drive data collection to sensor simulation, supporting both object-level and sensor-level validation with reduced data input requirements.

Keywords: Teleoperation Control Systems for Vehicles, Human Factors Analysis in Vehicle Design, Real-Time Control Strategies
Abstract: This study investigates disengagements of Remote Driving Systems (RDS) based on interventions by an in-vehicle Safety Drivers (SD) in real-world Operational Design Domains (ODD) with a focus on Remote Driver (RD) performance during their driving training. Based on an analysis of over 14,000 km on remote driving data, the relationship between the driving experience of 25 RD and the frequency of disengagements is systematically investigated. The results show that the number of SD interventions decreases significantly within the first 400 km of driving experience, which illustrates a clear learning curve of the RD. In addition, the most common causes for 183 disengagements analyzed are identified and categorized, whereby four main scenarios for SD interventions were identified and illustrated. The results emphasize the need for experience-based and targeted training programs aimed at developing basic driving skills early on, thereby increasing the safety, controllability and efficiency of RDS, especially in complex urban environment ODDs.

Keywords: Teleoperation Control Systems for Vehicles, Level 4-5 Autonomous Driving Systems Architecture
Abstract: Teleoperation is a key enabler for future mobility, supporting Automated Vehicles in rare and complex scenarios beyond the capabilities of their automation. Despite ongoing research, no open source software currently combines Remote Driving, e.g., via steering wheel and pedals, Remote Assistance through high-level interaction with automated driving software modules, and integration with a real-world vehicle for practical testing. To address this gap, we present a modular, open source teleoperation software stack that can interact with an automated driving software, e.g., Autoware, enabling Remote Assistance and Remote Driving. The software features standardized interfaces for seamless integration with various real-world and simulation platforms, while allowing for flexible design of the human-machine interface. The system is designed for modularity and ease of extension, serving as a foundation for collaborative development on individual software components as well as realistic testing and user studies. To demonstrate the applicability of our software, we evaluated the latency and performance on different vehicle platforms in simulation and real-world. The source code is available on GitHub.

Keywords: Human Factors Analysis in Vehicle Design, Teleoperation Control Systems for Vehicles
Abstract: Teleoperation is considered an interim solution for the introduction of automated vehicles in public transport. It describes the possibility for a remote driver to take control of the vehicle if there are uncertainties regarding the current driving situation. Despite the efforts of researchers and legislators and the obvious advantages of the technology, explicit regulations for teleoperation are still pending. One reason for this is the absence of clarity regarding which aspects need to be regulated and in what form. The aim of this paper is to provide an overview of different categories and specifications that need to be defined to support the derivation of legal requirements. For this purpose, we conducted a workshop in Germany with experts from various disciplines who have been involved in the integration of automated driving for many years. Based on insights from the fields of law, technology, human factors, and licensing authorities, a list of the requirements was developed. By presenting a holistic approach, the results should help to develop a common consensus on the scope of the regulations and to define clear requirements and specifications for the approval process of teleoperation on public roads.