In autonomous driving trajectory prediction, it is important to generate multi-modal trajectories. As a generative method, diffusion model has been increasingly adopted in the field of trajectory prediction. In this paper, we propose CDJMP (Conditional Diffusion model-based Joint Motion Prediction), an adaptive conditional diffusion framework designed for multi-agent trajectory forecasting in autonomous driving. However, when conventional diffusion models generate multi-modal trajectories, particularly for multi-agent joint prediction, they often fail to balance the diversity of generated trajectories and the accuracy of prediction. At the same time, diffusion models also suffer from time-consuming inference. To address these issues, CDJMP utilizes a two-stage framework to generate diverse and highly accurate trajectories. In the first stage, a probabilistic initializer predicts proposal trajectories together with adaptive denoising steps. In the second stage, a group-aware conditional encoder captures dynamic multi-agent interactions and guides the diffusion process to produce coherent multimodal outcomes. Experiments on the INTERACTION and Argoverse datasets demonstrate that CDJMP achieves state-of-the-art performance, reducing minADE and minFDE by up to 9
Although the majority of Intelligent Transportation Systems (ITS) and risk-prediction frameworks keep considering pavement quality and dynamic traffic behavior as separate phenomena, road infrastructure degradation and traffic flow instability both contribute to risky driving situations. This study combines segment-level Pavement Condition Index (PCI) data with actual traffic observations from New York City to present an integrated, data-driven methodology for simulating infrastructure-induced unsafe driving circumstances. A supervised machine learning model is developed by combining measures of traffic congestion, speed variation, and pavement deterioration to estimate hourly instability risk, which serves as a proxy for risky driving behavior. Evaluating on a temporally separated test set, the infrastructure-aware model achieves an ROC-AUC of 0.9804 and a PR-AUC of 0.9074, outperforming a traffic-only baseline (ROC-AUC 0.8785, PR-AUC 0.5833). These instability indicators correspond to high-level behavioral patterns commonly observed in ITS monitoring contexts, without relying on raw visual data. Model explainability using SHAP indicates that pavement condition and congestion are the most influential features, with comparable contributions to instability prediction. Predicted risk probabilities are geospatially mapped to identify infrastructure-driven hotspots, and the ORQCIAM framework demonstrates how such risk outputs can inform infrastructure-aware routing and maintenance prioritization. The findings reveal that machine learning enhanced with pavement condition data offers a data-driven approach for predicting hazardous driving situations and supporting infrastructure-aware decision-making, demonstrating how infrastructure-aware risk estimates might help with routing analysis and repair priority in future ITS applications.
Large Language Models (LLMs) offer powerful reasoning capabilities for industrial process control, yet their non-deterministic nature, susceptibility to hallucination, and lack of intrinsic physical understanding make direct deployment in safety-critical environments unacceptable. This paper addresses five research questions on safely integrating LLM-based reasoning into industrial process automation through the Autonomous Action Execution (AAE) framework. For safe architectural integration (RQ1), we present a four-layer multi-agent architecture that confines LLM inference to an observation-only Monitor layer while safety-critical decisions are made by deterministic Verification and Execution agents. For structuring heterogeneous plant data (RQ2), we introduce a text-level aggregation framework with pluggable analyzers that transforms SCADA states, time-series measurements, Piping and Instrumentation Diagrams (P IDs), and Standard Operating Procedures (SOPs) into contextually rich documents for LLM consumption. For automated validation (RQ3), a P ID-grounded method uses graph traversal over the P ID topology to verify physical consistency of LLM-generated proposals, checking tag existence, actuatability, fail-state consistency, and downstream impact. For quantifiable context enrichment (RQ4), a graduated baseline comparison (B0–B3) demonstrates the incremental value of each pipeline component. For cross-domain generalisability (RQ5), evaluation across five industrial scenarios—three derived from the Tennessee Eastman Process (TEP) benchmark (Downs Vogel, 1993) providing community-standard validation, plus two retained scenarios (PolyReactor, Dryer) that establish performance boundaries from best-case (zero hallucination) to worst-case (70
Deep learning-based autonomous inspection of power grid insulators is challenged by data imbalance and model opacity. This paper presents an end-to-end solution integrating advanced data synthesis, detection, classification, and explainability. First, a conditional diffusion model generates realistic synthetic fault images to balance the dataset. A two-stage architecture based on You Only Look Once version 26 (YOLO26) extra-large and Shifted windows (Swin)-V2-B, called YOLO26-Swin, fine-tuned with Bayesian optimization, performs robust insulator detection and then fault classification. Finally, a novel SHapley Additive exPlanations with Class Activation Mapping (SHAP-CAM) method provides intuitive visual explanations for model predictions. Extensive experiments validate our framework’s superiority: it achieves an F1-score of 0.98149 and a mean Average Precision (mAP)@[0.5] of 0.98951, exceeding leading detection and classification models. This work highlights the efficacy of diffusion models for data augmentation in critical infrastructure and advances the interpretability of vision-based inspection systems.
The electric power system is the largest man-made autonomous intelligent machine, which is of vital importance but vulnerable to cyber threats. This study investigates cyber-attack-induced equilibria, i.e., steady-state operation points, of two-terminal high-voltage direct-current (HVDC) power transmission systems, which play a critical role in maintaining power balances. By analytically mapping the control curves of two HVDC converters onto a common control plane, we show that originally linear characteristics become quadratic or more complex. Then, we derive propositions characterizing attack-induced switching of HVDC control strategies as well as the consequent shift of HVDC equilibrium points. Time-domain MATLAB/Simulink case studies validate the theoretical results, highlighting abrupt bulk power reversals and bifurcation-like behaviors under small measurement perturbations. The proposed framework provides HVDC system operators with explicit tools to assess and mitigate attack-driven HVDC operating risks.
