arXiv subjects

Jiarong Wei

Publications and source records attributed to Jiarong Wei.

6 recordsLinked to original sources

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

CounterPlay: Counterfactual Post-Training for Self-Play Driving Policies

Self-play in high-throughput simulators yields driving policies with robust closed-loop performance, but improvement per unit of simulation diminishes as training scales. Policies learn to handle common situations early, while further rollouts repeatedly encounter unresolved failures. Post-training offers an opportunity to target these failures, but existing methods primarily evaluate alternative actions or continuations at visited states, although successful recovery may require changing driving style earlier. We propose CounterPlay, a counterfactual self-play post-training approach that backtracks from failed tasks and retries them under alternate driving styles. CounterPlay rests on three key components. First, failure-driven backtracking uses the policy's value estimates to select an earlier stored state from which to retry the task. Second, reward conditioning enables a single policy to retry the task from this state using candidate styles ranging from cautious to aggressive. Third, CounterPlay retains task-completing retries only if no other vehicle incurs a new or earlier collision or off-road event relative to the factual branch. Retries that pass verification with fresh randomness are then distilled into the policy under its deployment condition. On BehaviorBench, CounterPlay achieves state-of-the-art scores on both the Interactive and Random splits across all eight traffic regimes using 1B post-training transitions, which is just 1% of the anchor's 100B self-play training budget. Improvements over the anchor hold across all three evaluated driving styles. CounterPlay resolves a substantial fraction of the anchor's timeout cases on BehaviorBench and achieves a balance between task completion and safety that neither continued self-play nor adopting a more aggressive driving style attains.

cs.RO↗

arXiv2026

CoPark: Learning Reactive Parking via Self-Play

Learning a single policy that reaches a goal with high geometric precision while interacting safely with nearby agents poses conflicting objectives. Precision favors commitment to a fixed geometric plan, whereas interaction requires immediate deviation when another agent intrudes, causing policies optimized for one objective to often fail at the other. We study this problem in the context of reactive autonomous parking, where multiple vehicles must reach assigned slots with sub-meter terminal accuracy while remaining responsive to neighboring vehicles throughout the maneuver. We propose CoPark, a multi-agent self-play RL approach built on a residual-policy architecture. A precomputed offline plan provides a fixed action prior, while a residual head learns the reactive corrections. The residual policy learns behaviors under self-play, where data and scripting fall short, while the fixed prior holds the slot-frame geometry that pure policies struggle to reach reliably. The key design is a partner-threat-modulated, channel-asymmetric release of the prior. A continuous threat signal shifts authority of the longitudinal channel to the residual head to enable yielding, while the lateral channel remains anchored to the precomputed reference to preserve sub-meter slot alignment. A closed-loop refinement layer corrects residual terminal error from action-grid discretization. We train our policy on six parking lots and evaluate zero-shot on our new reactive-parking benchmark spanning Dragon Lake Parking (DLP) and DeepScenario Open 3D (DSC3D). CoPark achieves ~70-85% success with only 3-6% collision rate, substantially outperforming classical, imitation-learning, and large-scale RL baselines. Importantly, the results demonstrate emergent interaction behaviors such as reverse-yielding, mid-maneuver yielding, tight-corridor passing, and queuing.

cs.RO↗

arXiv2026

ParkDiffusion++: Ego Intention Conditioned Joint Multi-Agent Trajectory Prediction for Automated Parking using Diffusion Models

Automated parking is a challenging operational domain for advanced driver assistance systems, requiring robust scene understanding and interaction reasoning. The key challenge is twofold: (i) predict multiple plausible ego intentions according to context and (ii) for each intention, predict the joint responses of surrounding agents, enabling effective what-if decision-making. However, existing methods often fall short, typically treating these interdependent problems in isolation. We propose ParkDiffusion++, which jointly learns a multi-modal ego intention predictor and an ego-conditioned multi-agent joint trajectory predictor for automated parking. Our approach makes several key contributions. First, we introduce an ego intention tokenizer that predicts a small set of discrete endpoint intentions from agent histories and vectorized map polylines. Second, we perform ego-intention-conditioned joint prediction, yielding socially consistent predictions of the surrounding agents for each possible ego intention. Third, we employ a lightweight safety-guided denoiser with different constraints to refine joint scenes during training, thus improving accuracy and safety. Fourth, we propose counterfactual knowledge distillation, where an EMA teacher refined by a frozen safety-guided denoiser provides pseudo-targets that capture how agents react to alternative ego intentions. Extensive evaluations demonstrate that ParkDiffusion++ achieves state-of-the-art performance on the Dragon Lake Parking (DLP) dataset and the Intersections Drone (inD) dataset. Importantly, qualitative what-if visualizations show that other agents react appropriately to different ego intentions.

cs.RO↗

arXiv2025

ParkDiffusion: Heterogeneous Multi-Agent Multi-Modal Trajectory Prediction for Automated Parking using Diffusion Models

Automated parking is a critical feature of Advanced Driver Assistance Systems (ADAS), where accurate trajectory prediction is essential to bridge perception and planning modules. Despite its significance, research in this domain remains relatively limited, with most existing studies concentrating on single-modal trajectory prediction of vehicles. In this work, we propose ParkDiffusion, a novel approach that predicts the trajectories of both vehicles and pedestrians in automated parking scenarios. ParkDiffusion employs diffusion models to capture the inherent uncertainty and multi-modality of future trajectories, incorporating several key innovations. First, we propose a dual map encoder that processes soft semantic cues and hard geometric constraints using a two-step cross-attention mechanism. Second, we introduce an adaptive agent type embedding module, which dynamically conditions the prediction process on the distinct characteristics of vehicles and pedestrians. Third, to ensure kinematic feasibility, our model outputs control signals that are subsequently used within a kinematic framework to generate physically feasible trajectories. We evaluate ParkDiffusion on the Dragon Lake Parking (DLP) dataset and the Intersections Drone (inD) dataset. Our work establishes a new baseline for heterogeneous trajectory prediction in parking scenarios, outperforming existing methods by a considerable margin.

cs.RO↗

arXiv2024

BaSAL: Size-Balanced Warm Start Active Learning for LiDAR Semantic Segmentation

Active learning strives to reduce the need for costly data annotation, by repeatedly querying an annotator to label the most informative samples from a pool of unlabeled data, and then training a model from these samples. We identify two problems with existing active learning methods for LiDAR semantic segmentation. First, they overlook the severe class imbalance inherent in LiDAR semantic segmentation datasets. Second, to bootstrap the active learning loop when there is no labeled data available, they train their initial model from randomly selected data samples, leading to low performance. This situation is referred to as the cold start problem. To address these problems we propose BaSAL, a size-balanced warm start active learning model, based on the observation that each object class has a characteristic size. By sampling object clusters according to their size, we can thus create a size-balanced dataset that is also more class-balanced. Furthermore, in contrast to existing information measures like entropy or CoreSet, size-based sampling does not require a pretrained model, thus addressing the cold start problem effectively. Results show that we are able to improve the performance of the initial model by a large margin. Combining warm start and size-balanced sampling with established information measures, our approach achieves comparable performance to training on the entire SemanticKITTI dataset, despite using only 5% of the annotations, outperforming existing active learning methods. We also match the existing state-of-the-art in active learning on nuScenes. Our code is available at: https://github.com/Tony-WJR/BaSAL.

cs.CV↗