arXiv subjects

J. Marius Zöllner

Publications and source records attributed to J. Marius Zöllner.

At least 19 recordsLinked to original sources

arXiv2026

Temporal Residual Bottleneck for Robust Asynchronous Collaborative Perception

Collaborative perception extends the sensing range of autonomous vehicles, but its performance degrades when shared features arrive stale or incomplete. Most latency-robust methods compensate delayed collaborator features through flow-guided alignment or direct feature transport. In this work, we formulate asynchronous collaborative perception as temporal residual prediction. Our Temporal Residual Bottleneck keeps a deterministic pose-warped collaborator feature as a conservative anchor and uses a $Δt$-conditioned xLSTM to extract residual temporal evidence from the available history. A detector-facing residual bottleneck then applies only gated, regularized corrections before ego-side fusion, reducing the risk of overwriting reliable static structure when temporal correspondence is uncertain. Experiments on DAIR-V2X and OPV2V show that our method is especially effective under severe fixed/irregular delays and packet drops. On DAIR-V2X, the reported checkpoint trades a small amount of synchronized peak accuracy for better robustness under stronger communication degradation. Controlled diagnostics further indicate that direct feature transport has oracle headroom but can become unreliable when deployed without accurate correspondence. These results support temporal residual fusion as a practical alternative for asynchronous and incomplete collaborative perception. Code will be publicly released at https://url.fzi.de/8dk38.

cs.CV↗

arXiv2026

Beyond Waypoint Regression: Query-Based Cost Learning over Reachable Ego Futures for End-to-End Driving

End-to-end planners based on waypoint regression achieve strong open-loop accuracy, but they primarily learn to mimic expert geometry and remain difficult to adapt to deployment-time safety constraints. We propose a query-based cost-learning framework that estimates bounded costs for dynamically reachable ego trajectory queries, rather than dense BEV cells or a small regressed trajectory set. Compact joint scene tokens capture coherent multimodal agent futures, while contingency-aware cost aggregation and cost-guided intra-cluster MPPI mixing convert the learned cost topology into feasible ego plans. On nuScenes, our method improves over prior cost-estimation planners such as ST-P3 and NMP, outperforms most regression baselines in collision rate, while remaining competitive in L2, and retaining an interpretable cost interface. On real-world driving logs, the proposed planner reduces collision rates compared with SparseDrive and Alpamayo without fine-tuning, while maintaining a diverse set of candidate trajectories.

cs.RO↗

arXiv2026

Dense Coverage, Sparse Refinement: Byte-Constrained Cooperative Perception

Collaborative perception improves autonomous perception by sharing intermediate Bird's-Eye-View (BEV) features across connected agents, but dense feature exchange is difficult to deploy under strict Vehicle-to-Everything (V2X) bandwidth limits. Existing efficient methods typically either compress the full feature map uniformly, spending bits on low-value background, or sparsify communication, risking the loss of useful context. We propose a coverage-refinement design for byte-constrained cooperative perception: each agent transmits a highly compressed coarse layer over the full BEV map and allocates the remaining budget to selected high-resolution patches. A Task-Aware Benefit Selector ranks cells by estimated downstream utility, enabling deterministic budgeted refinement and zero-retraining adaptation to changing bandwidth. The receiver reconstructs a dense BEV tensor compatible with standard fusion modules. Experiments on DAIR-V2X and OPV2V show strong accuracy-payload trade-offs at kilobyte-scale budgets. On DAIR-V2X, our method reaches 0.60 AP@0.7 at only 1.87 KB per non-ego agent, compared with 0.52 at 4.61 KB for uniform SimVQ compression. Controlled diagnostics further show that the gain arises from coverage-refinement allocation rather than quantization alone. Code will be published.

cs.CV↗

arXiv2026

Safety-Critical Scenarios Emerge from Initial Scenes

Safety-critical driving scenario generation has largely focused on manipulating the behavior of surrounding agents while starting from an initial scene from driving data. This assumption can limit the space of discoverable failures, since driving data can provide little opportunity for meaningful interaction. For example, in the Waymo Open Motion Dataset, 20.44% of recorded slices feature a stationary ego vehicle that never moves, and 30.39% of initial frames contain no nearby traffic participants within 10 meters. We instead study safety-critical scenario generation as an initialization problem: given agnostic black-box driving policies, we learn to generate realistic initial scenes that are more likely to evolve into critical interactions. We propose AdvScene, a conditional latent diffusion model that is trained in two stages. Starting from pretraining on naturalistic driving data, we post-train the adversarial-agent generation branch using reinforcement learning with feedback from closed-loop simulator rollouts. Conditioning on ego driving displacement prevents the ego from remaining static, and RL finetuning induces criticality directly with non-differentiable safety-critical metrics. Experiments on the Waymo dataset across 12 combinations of ego and traffic policies show that our AdvScene substantially increases the rate of ego-fault collision events and TTC<3s events.

cs.RO↗

arXiv2026

User Experience in Human-Machine Interaction: Insights from Field Studies in Autonomous Mobility

Autonomous vehicles (AVs) promise safer, cleaner, and more inclusive mobility, yet large-scale adoption is hindered by user acceptance rather than by technical challenges. Prior studies on acceptance and user experience largely rely on surveys, simulators or Wizard-of-Oz setups, often over-representing technologically enthusiastic participants and focusing on drivers instead of passengers. We address this gap with real-world field studies with AVs in real traffic, totaling 144 participants. Using multi-modal sensing, we evaluated EGG, heartbeat, breathing, camera and voice signals for affect inference in combination with vehicle data. Our results show that breathing, camera and voice measurements are reliable and pratical in naturalistic passenger contexts. We further contribute a validated study protocol, a self-assessment app for real-time assessment during human-machine interaction, and a tailored questionnaire to capture participant attitudes towards AVs. By grounding UX evaluation in real-world contexts, this work lays a foundation for user-centered design of autonomous mobility systems and robotics in general. Our work bridges the gap between affective computing and technical implementation of autonomous vehicles.

cs.HC↗

arXiv2026

Uncertainty-Aware and Temporally Regulated Expert Advice in Reinforcement Learning for Autonomous Driving

Exploration in reinforcement learning for autonomous driving is inherently unsafe: agents must experience novel behaviors to learn, yet exploration can lead to collisions or off-road driving. We propose an uncertainty-aware framework that leverages expert advice to guide exploration while avoiding long-term dependence. Advice is triggered when epistemic or aleatoric uncertainty exceeds adaptive thresholds derived from rolling buffers, ensuring advice evolves with the agent's confidence. A commitment-cooldown strategy with a stochastic early-stop heuristic regulates the duration and frequency of guidance, exposing the agent to coherent maneuvers without exhausting the advice budget. Expert and agent experiences are combined in a shared replay buffer within an off-policy implicit quantile network (IQN) backbone, enabling efficient reuse of expert trajectories. Experiments in CARLA show that our method outperforms the IQN baseline, improving success by 5-7% and reducing failures, demonstrating that risk-sensitive uncertainty coupled with regulated expert integration enables safer and more efficient exploration for sensor-based RL policy learning in unsignalized intersection navigation.

cs.AI↗

arXiv2026

Segment to Focus: Guiding Latent Action Models in the Presence of Distractors

Latent action models (LAMs) offer a promising path to pre-training embodied agents on large amounts of action-free video. They infer latent actions between consecutive observations that can later be decoded to ground-truth actions using a small number of labels. However, recent work has shown that this recipe fails in the presence of action-correlated visual distractors common in real-world video, such as dynamic backgrounds, camera shake, or other moving objects. In these scenarios, the standard reconstruction objective drives latent actions to encode exogenous motion instead of agent-controlled dynamics, resulting in policies that underperform when fine-tuned. We observe, however, that endogenous and exogenous factors are typically spatially separated in pixel space: control-relevant change is concentrated on the agent, while distractor motion occurs elsewhere. We exploit this observation by restricting the reconstruction objective to agent pixels, forcing latent actions to explain agent-controlled dynamics rather than exogenous ones. We call this method MaskLAM; it obtains the agent mask zero-shot from off-the-shelf segmentation foundation models (e.g., SAM) and requires no architectural changes, auxiliary losses, or action labels during pre-training. Across two continuous-control benchmarks (Distracting Control Suite, Distracting Meta-World), MaskLAM reduces normalized linear-probe MSE by up to $3.51\times$ and improves normalized return by up to $4.97\times$ over LAPO, while narrowing the gap to LAOM-Labels, which relies on ground-truth action supervision.

cs.LG↗

arXiv2026

CIG: Exploration via Conditional Information Gain

Intrinsic rewards for exploration in reinforcement learning condition on different contexts: lifelong rewards score each transition against accumulated experience but ignore within-rollout redundancy; episodic rewards penalize intra-trajectory repetition but discard lifetime progress. Hybrid methods combine both signals through heuristic weights or require Gaussian-process dynamics that do not scale beyond low-dimensional state spaces. Trajectory-level information gain decomposes into per-step terms that condition on the replay buffer and rollout prefix simultaneously, but remains intractable for deep models. We derive the Conditional Information Gain (CIG) reward as a tractable surrogate: a log-determinant objective over an ensemble disagreement kernel whose Cholesky factorization yields causal per-step rewards that retain both conditioning sets while scaling to high-dimensional state spaces. We instantiate CIG in a model-based setting, where rollouts are short and within-rollout corrections remain largely unexplored. Across twelve tasks spanning discrete (MiniGrid) and continuous control (OGBench), in both clean and stochastic-distractor settings, CIG outperforms or matches prior exploration methods while remaining robust to stochastic distractors.

cs.LG↗

arXiv2026

Recall to Predict: Grounding Motion Forecasting in Interpretable Motion Bank

Motion forecasting often requires trading interpretability for predictive accuracy. Standard anchor-based architectures rely on opaque latent queries that are highly prone to latent collapse, or naive trajectory sampling that limits multi-modal diversity. We propose an end-to-end differentiable framework that grounds predictions in a comprehensive "motion bank", a structured embedding space of physically realizable trajectories constructed via contrastive learning. Rather than regressing paths from a blank slate, our architecture dynamically retrieves explicit motion priors using a novel Anchor Retrieval Layer. This module adapts orthogonally initialized queries via a Dual-Level Gated Cross-Attention mechanism and executes discrete trajectory selection using a Straight-Through Gumbel-Softmax estimator to preserve continuous gradient flow. The retrieved semantically grounded anchors are then geometrically refined by a DETR-style decoder, optimized jointly with a Winner-Takes-All (WTA) kinematic Gaussian Mixture Model (GMM), a latent diversity penalty, and a soft-min weighted endpoint loss. By strictly conditioning the decoding phase on diverse, interpretable motion primitives, our approach eliminates the "black box" of standard latent queries while achieving competitive multi-modal accuracy on the Argoverse 2 and Waymo Open Motion datasets. Code is available at: https://github.com/abviv/recall2predict

cs.CV↗

arXiv2026

Towards a Systematic Risk Assessment of Deep Neural Network Limitations in Autonomous Driving Perception

Safety and security are essential for the admission and acceptance of automated and autonomous vehicles. Deep neural networks (DNNs) are widely used for perception and further components of the autonomous driving (AD) stack. However, they possess several limitations, including lack of generalization, efficiency, explainability, plausibility, and robustness. These insufficiencies can pose significant risks to autonomous driving systems. However, hazards, threats, and risks associated with DNN limitations in this domain have not been systematically studied so far. In this work, we propose a joint workflow for risk assessment combining the hazard analysis and risk assessment (HARA) following ISO 26262 and threat analysis and risk assessment (TARA) following the ISO/SAE 21434 to identify and analyze risks arising from inherent DNN limitations in AD perception.

cs.CR↗

arXiv2026

Domain-Specialized Object Detection via Model-Level Mixtures of Experts

Mixture-of-Experts (MoE) models provide a structured approach to combining specialized neural networks and offer greater interpretability than conventional ensembles. While MoEs have been successfully applied to image classification and semantic segmentation, their use in object detection remains limited due to challenges in merging dense and structured predictions. In this work, we investigate model-level mixtures of object detectors and analyze their suitability for improving performance and interpretability in object detection. We propose an MoE architecture that combines YOLO-based detectors trained on semantically disjoint data subsets, with a learned gating network that dynamically weights expert contributions. We study different strategies for fusing detection outputs and for training the gating mechanism, including balancing losses to prevent expert collapse. Experiments on the BDD100K dataset demonstrate that the proposed MoE consistently outperforms standard ensemble approaches and provides insights into expert specialization across domains, highlighting model-level MoEs as a viable alternative to traditional ensembling for object detection. Our code is available at https://github.com/KASTEL-MobilityLab/mixtures-of-experts/.

cs.CV↗

arXiv2026

Design and Behavior of Sparse Mixture-of-Experts Layers in CNN-based Semantic Segmentation

Sparse mixture-of-experts (MoE) layers have been shown to substantially increase model capacity without a proportional increase in computational cost and are widely used in transformer architectures, where they typically replace feed-forward network blocks. In contrast, integrating sparse MoE layers into convolutional neural networks (CNNs) remains inconsistent, with most prior work focusing on fine-grained MoEs operating at the filter or channel levels. In this work, we investigate a coarser, patch-wise formulation of sparse MoE layers for semantic segmentation, where local regions are routed to a small subset of convolutional experts. Through experiments on the Cityscapes and BDD100K datasets using encoder-decoder and backbone-based CNNs, we conduct a design analysis to assess how architectural choices affect routing dynamics and expert specialization. Our results demonstrate consistent, architecture-dependent improvements (up to +3.9 mIoU) with little computational overhead, while revealing strong design sensitivity. Our work provides empirical insights into the design and internal dynamics of sparse MoE layers in CNN-based dense prediction. Our code is available at https://github.com/KASTEL-MobilityLab/moe-layers/.

cs.CV↗

arXiv2026

Diverse and Adaptive Behavior Curriculum for Autonomous Driving: A Student-Teacher Framework with Multi-Agent RL

Autonomous driving faces challenges in navigating complex real-world traffic, requiring safe handling of both common and critical scenarios. Reinforcement learning (RL), a prominent method in end-to-end driving, enables agents to learn through trial and error in simulation. However, RL training often relies on rule-based traffic scenarios, limiting generalization. Additionally, current scenario generation methods focus heavily on critical scenarios, neglecting a balance with routine driving behaviors. Curriculum learning, which progressively trains agents on increasingly complex tasks, is a promising approach to improving the robustness and coverage of RL driving policies. However, existing research mainly emphasizes manually designed curricula, focusing on scenery and actor placement rather than traffic behavior dynamics. This work introduces a novel student-teacher framework for automatic curriculum learning. The teacher, a graph-based multi-agent RL component, adaptively generates traffic behaviors across diverse difficulty levels. An adaptive mechanism adjusts task difficulty based on student performance, ensuring exposure to behaviors ranging from common to critical. The student, though exchangeable, is realized as a deep RL agent with partial observability, reflecting real-world perception constraints. Results demonstrate the teacher's ability to generate diverse traffic behaviors. The student, trained with automatic curricula, outperformed agents trained on rule-based traffic, achieving higher rewards and exhibiting balanced, assertive driving.

cs.RO↗

arXiv2026

Bridging Simulation and Usability: A User-Friendly Framework for Scenario Generation in CARLA

Autonomous driving promises safer roads, reduced congestion, and improved mobility, yet validating these systems across diverse conditions remains a major challenge. Real-world testing is expensive, time-consuming, and sometimes unsafe, making large-scale validation impractical. In contrast, simulation environments offer a scalable and cost-effective alternative for rigorous verification and validation. A critical component of the validation process is scenario generation, which involves designing and configuring traffic scenarios to evaluate autonomous systems' responses to various events and uncertainties. However, existing scenario generation tools often require programming knowledge, limiting accessibility for non-technical users. To address this limitation, we present an interactive, no-code framework for scenario generation. Our framework features a graphical interface that enables users to create, modify, save, load, and execute scenarios without needing coding expertise or detailed simulation knowledge. Unlike script-based tools such as Scenic or ScenarioRunner, our approach lowers the barrier to entry and supports a broader user base. Central to our framework is a graph-based scenario representation that facilitates structured management, supports both manual and automated generation, and enables integration with deep learning-based scenario and behavior generation methods. In automated mode, the framework can randomly sample parameters such as actor types, behaviors, and environmental conditions, allowing the generation of diverse and realistic test datasets. By simplifying the scenario generation process, this framework supports more efficient testing workflows and increases the accessibility of simulation-based validation for researchers, engineers, and policymakers.

cs.RO↗

arXiv2026

Beyond Scalar Rewards: Distributional Reinforcement Learning with Preordered Objectives for Safe and Reliable Autonomous Driving

Autonomous driving involves multiple, often conflicting objectives such as safety, efficiency, and comfort. In reinforcement learning (RL), these objectives are typically combined through weighted summation, which collapses their relative priorities and often yields policies that violate safety-critical constraints. To overcome this limitation, we introduce the Preordered Multi-Objective MDP (Pr-MOMDP), which augments standard MOMDPs with a preorder over reward components. This structure enables reasoning about actions with respect to a hierarchy of objectives rather than a scalar signal. To make this structure actionable, we extend distributional RL with a novel pairwise comparison metric, Quantile Dominance (QD), that evaluates action return distributions without reducing them into a single statistic. Building on QD, we propose an algorithm for extracting optimal subsets, the subset of actions that remain non-dominated under each objective, which allows precedence information to shape both decision-making and training targets. Our framework is instantiated with Implicit Quantile Networks (IQN), establishing a concrete implementation while preserving compatibility with a broad class of distributional RL methods. Experiments in Carla show improved success rates, fewer collisions and off-road events, and deliver statistically more robust policies than IQN and ensemble-IQN baselines. By ensuring policies respect rewards preorder, our work advances safer, more reliable autonomous driving systems.

cs.RO↗

arXiv2026

Balancing Progress and Safety: A Novel Risk-Aware Objective for RL in Autonomous Driving

Reinforcement Learning (RL) is a promising approach for achieving autonomous driving due to robust decision-making capabilities. RL learns a driving policy through trial and error in traffic scenarios, guided by a reward function that combines the driving objectives. The design of such reward function has received insufficient attention, yielding ill-defined rewards with various pitfalls. Safety, in particular, has long been regarded only as a penalty for collisions. This leaves the risks associated with actions leading up to a collision unaddressed, limiting the applicability of RL in real-world scenarios. To address these shortcomings, our work focuses on enhancing the reward formulation by defining a set of driving objectives and structuring them hierarchically. Furthermore, we discuss the formulation of these objectives in a normalized manner to transparently determine their contribution to the overall reward. Additionally, we introduce a novel risk-aware objective for various driving interactions based on a two-dimensional ellipsoid function and an extension of Responsibility-Sensitive Safety (RSS) concepts. We evaluate the efficacy of our proposed reward in unsignalized intersection scenarios with varying traffic densities. The approach decreases collision rates by 21\% on average compared to baseline rewards and consistently surpasses them in route progress and cumulative reward, demonstrating its capability to promote safer driving behaviors while maintaining high-performance levels.

cs.RO↗

arXiv2026

Automatic Curriculum Learning for Driving Scenarios: Towards Robust and Efficient Reinforcement Learning

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 enhanced 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.

cs.RO↗

arXiv2026

Boundary-Guided Trajectory Prediction for Road Aware and Physically Feasible Autonomous Driving

Accurate prediction of surrounding road users' trajectories is essential for safe and efficient autonomous driving. While deep learning models have improved performance, challenges remain in preventing off-road predictions and ensuring kinematic feasibility. Existing methods incorporate road-awareness modules and enforce kinematic constraints but lack plausibility guarantees and often introduce trade-offs in complexity and flexibility. This paper proposes a novel framework that formulates trajectory prediction as a constrained regression guided by permissible driving directions and their boundaries. Using the agent's current state and an HD map, our approach defines the valid boundaries and ensures on-road predictions by training the network to learn superimposed paths between left and right boundary polylines. To guarantee feasibility, the model predicts acceleration profiles that determine the vehicle's travel distance along these paths while adhering to kinematic constraints. We evaluate our approach on the Argoverse-2 dataset against the HPTR baseline. Our approach shows a slight decrease in benchmark metrics compared to HPTR but notably improves final displacement error and eliminates infeasible trajectories. Moreover, the proposed approach has superior generalization to less prevalent maneuvers and unseen out-of-distribution scenarios, reducing the off-road rate under adversarial attacks from 66% to just 1%. These results highlight the effectiveness of our approach in generating feasible and robust predictions.

cs.RO↗