Papers for

automated vehicle developers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Learned trajectory prediction keeps an edge at low update rates

Learned versus Kinematic Trajectory Prediction under Reduced Observation Frequency

Abstract: Trajectory prediction models for automated driving are usually trained on data with a fixed sampling frequency, while positions received via vehicle-to-everything (V2X) communication arrive at 1-10 Hz. We study whether a learned predictor that uses only the observation history keeps its advantage over constant velocity (CV) and constant acceleration (CA) models when the history frequency drops. On the Argoverse 2 motion forecasting dataset, the 5 s history of the focal agent was downsampled from 10 Hz to 5, 2, and 1 Hz, and the 6 s future was predicted without maps or neighboring agents. An LSTM-based recurrent neural network (RNN) trained for each frequency was compared with the same network trained at 10 Hz and applied to sparse histories after linear interpolation. Hypotheses and thresholds were fixed before training, the main RNNs were trained with ten seeds each, and results on 24,984 test scenarios have bootstrap confidence intervals over scenarios and seeds. The average displacement error of the frequency-specific RNN grew by 3.1% at 5 Hz, 10.1% at 2 Hz, and 20.3% at 1 Hz. Contrary to our hypothesis, its advantage over the better kinematic model did not vanish on straight motion and was 24-30% on both straight motion and turns at every frequency, while for pedestrians CV was more accurate. The 10 Hz model with interpolation was at most 1.3% less accurate than the frequency-specific model down to 2 Hz but 11.5% less accurate at 1 Hz. Each hypothesis was only partly supported.

Tue 17 NovMachine LearningRobotics
The gist
Automated driving systems predict where vehicles will move based on recent position data. This study looked at how well a learned prediction model works when it gets position updates less often, like every second instead of ten times per second. The authors found that the learned model still predicts better than simple motion models, even at lower update frequencies, especially for vehicles moving straight or turning. However, for pedestrians, simple models often did better. Using training for each data frequency worked better than just tweaking a high-frequency model.
Open → 2611.23198v1

Planning safely inside learned models with convex optimization

Successive Convexification in Learned World Models for Safe Offline Planning

Abstract: World models learned from offline data allow an agent to plan without further interaction with the environment, but the sampling-based planners used with them handle safety constraints only by scoring or rejecting sampled action sequences. We instead pose safe planning in the latent space of a learned world model as a successive convex program (SCP), in which a learned discrete-time control barrier function is imposed as a constraint at every step of the horizon. SCP requires smooth dynamics, so we build the networks from smoothed MaxMin activations and spectrally normalised layers, and bound the resulting linearisation error in terms of the Lipschitz bound and the smoothing width. As SCP is a local method, we start it from a sampled plan. On two Safety-Gymnasium navigation tasks with offline data, this planner has fewer violations than MPPI and CEM drawing 128 times as many samples at equal wall-clock time in our main comparison, though not significantly, and earns greater return at longer horizons, and is much safer than SCP without a sampled start. SCP converged more often on Lipschitz-bounded models. Unlike sampling, it also reports at each step whether its plan is a feasible, first-order optimal KKT point of the planning problem in the model, or that no feasible plan was found nearby. Our implementation is publicly available at https://github.com/youngers2006/scwm.

Sat 14 NovMachine LearningRobotics
The gist
Planning actions safely using data from past experiences is hard because approaches often guess many possible plans and reject unsafe ones, which can be inefficient. The authors propose a new method that turns this into a step-by-step mathematical problem with safety rules built in. They smooth the model’s behavior to make this step-by-step approach work well and start it from a guess to find safer plans faster. In tests with simulated navigation tasks, their method found safer paths with fewer rule violations and gave a clearer answer if no safe plan was nearby.
Open → 2611.19466v1

Driving scene model improves reasoning and segmentation accuracy

DSRS: Teaching MLLMs to Perceive, Reason, and Segment in Complex Driving Scenes

Abstract: Existing reasoning segmentation frameworks typically employ multimodal large language models (MLLMs) with broad world knowledge to extract semantics in a human-like manner and generate corresponding segmentation masks. Nevertheless, these frameworks lack the necessary scene-specific awareness. This paper introduces driving scene reasoning segmentation (DSRS), a scene-specific task that requires models to interpret implicit instructions, infer spatial relations, detect potential risks, handle challenging conditions, and comprehend underlying traffic rules, ultimately generating accurate segmentation masks. Meanwhile, we establish a benchmark dataset to facilitate future research on the DSRS task. To pioneer this task, we propose DSRS-Net, a novel architecture with advanced reasoning segmentation capability for automated driving. Specifically, DSRS-Net seamlessly infuses geometric features for deeper reasoning segmentation, without compromising the original multimodal fusion mechanism. This strategy effectively mitigates the modality representation mismatch in MLLMs caused by RGB-only pretraining, while enhancing the integration of geometric, contextual, and linguistic features. Extensive experiments demonstrate the superior performance of DSRS-Net in both driving scene reasoning segmentation and other language-driven segmentation tasks across various public datasets. The source code and dataset will be released upon publication.

Wed 4 NovComputer Vision and Pattern Recognition
The gist
Understanding complex driving scenes is hard because it involves recognizing many things and their relationships, such as other cars, traffic rules, and risks. The authors introduce a new task called driving scene reasoning segmentation that challenges models to do all this and create detailed visual maps called segmentation masks. They built a special dataset to help train and test these models and designed DSRS-Net, a model that combines images with 3D information for better understanding. Their model performs better than others on this task and related language-guided tasks.
Open → 2611.06160v1

Hybrid model improves safety by combining braking and steering actions

Combining Evasive and Braking Reactions for Safety Reference Models in Automated Vehicles

Abstract: Computational models of careful and competent human drivers are essential for scenario-based evaluation of automated driving systems (ADS). However, most existing safety reference models primarily focus on longitudinal braking, neglecting the role of evasive steering in human collision avoidance. This paper proposes a hybrid Fuzzy-Safety Model (FSM-H) that integrates longitudinal mitigation and lateral avoidance within a unified behavioral framework. The braking component is governed by Proactive Fuzzy Safety (PFS) metrics, representing the erosion of longitudinal safety margins, while the steering component is driven by Criticality Fuzzy Safety for lane-change (CFS-LC), capturing lateral conflict severity and maneuver feasibility. A finite-state architecture models the sequential escalation from nominal driving to braking and, when necessary, to evasive steering, incorporating perception-reaction time and lane-check delays to reflect human decision processes. The model is evaluated in reconstructed high-criticality cut-in scenarios and compared with braking-only and steering-only reference strategies. Results show that the hybrid approach expands the preventability envelope while maintaining behavioral plausibility and computational tractability. The proposed framework provides a transparent and explainable human reference model suitable for simulation-based ADS safety benchmarking and regulatory assessment.

Thu 24 SeptRobotics
The gist
Automated cars need to understand how human drivers avoid crashes, but most models only focus on braking and ignore steering to dodge danger. The authors created a hybrid model that mixes braking and steering reactions based on how risky the situation is, just like a careful human driver would do. Their model includes realistic human reaction times and tests better in tricky traffic scenarios compared to models that only brake or only steer. This helps make simulated tests of self-driving cars safer and more realistic.
Open → 2609.29738v1

Pedestrian model mimics human crossing in complex traffic

A Human-Like Pedestrian Model for Automated Driving Simulations

Abstract: Automated vehicles must be able to interact with pedestrians safely and efficiently across diverse traffic situations. Although driving simulators offer a scalable testbed for learning such capabilities, existing theory-inspired pedestrian models are narrow in scope and limited to go/no-go crossing decisions in single-lane settings. While data-driven approaches can predict pedestrian behavior in complex situations, they lack sufficient observations in rare, safety-critical scenarios. Here, we propose an approach to training pedestrian models in simulators so that learned policies generate demonstrably human-like behavior in realistic, complex traffic scenarios, including multiple lanes, heavy traffic, and dangerous driving styles. Our technical contribution is a novel definition of pedestrian-vehicle interaction as a partially observable Markov decision process (POMDP) with theory-grounded perceptual, cognitive, and motor constraints. It accounts for the highly adaptive nature of human behavior in traffic and simulates how people adjust their responses according to perceived danger, time pressure, and the complexity of the situation. When trained via deep reinforcement learning (RL) with domain randomization in a simulator, the model reproduces the broadest range of empirical findings shown so far on human crossing behavior, including gap acceptance, yielding acceptance, hesitation, and evasive speed adjustment. We show that learned policies transfer to unseen traffic environments, and can be further adapted to local traffic norms with finetuning. Together, these results establish a blueprint for simulator-ready pedestrian models that can support the development and evaluation of automated driving systems.

Thu 24 SeptHuman-Computer Interaction
The gist
Automated cars need to understand how people cross streets safely in many different traffic situations. The authors created a computer model that learns to act like real pedestrians, even in tricky settings with multiple lanes and dangerous drivers. Their model thinks about what a pedestrian can see and how they decide when to cross, adjusting behavior for things like danger or hurry. When trained with lots of simulated experiences, it behaves like humans do, including how people hesitate or change speed to stay safe. This model can also be tweaked to match local traffic rules, helping test self-driving cars more realistically.
Open → 2609.29175v1