Robotic arms serve as critical actuation edge nodes in Industrial Internet of Things (IIoT)-enabled intelligent manufacturing systems. In distributed industrial cyber-physical architectures, multiple robotic manipulators are required to perform autonomous and cooperative motion planning under obstacle-rich and dynamically coupled environments. However, existing multi-agent deep reinforcement learning approaches often exhibit limited scalability, redundant observation processing, inefficient experience utilization, and unstable convergence when deployed in high-dimensional cooperative scenarios. To address these challenges, this paper proposes an Attention-based Prioritized Trajectory Multi-Agent Deep Deterministic Policy Gradient (ATP-MADDPG) framework tailored for edge-coordinated multi-arm systems in IIoT environments. The proposed framework incorporates an adaptive attention mechanism to selectively emphasize critical interaction features while suppressing redundant sensory information, thereby enhancing decision efficiency at distributed edge nodes. A prioritized sequence experience replay (PSER) strategy is further introduced to improve the utilization of cooperative trajectory data and accelerate policy evolution. In addition, curriculum learning is employed to enable progressive training and scalable policy refinement for complex multi-arm tasks. Extensive simulations demonstrate that, compared with conventional MADDPG and representative baselines, the proposed ATP-MADDPG achieves higher task success rates, improved cumulative rewards, faster convergence, and enhanced policy stability. These results validate the effectiveness of the proposed framework for distributed cooperative motion planning in IIoT-oriented robotic systems.

Text-based 3D human motion editing is a critical yet challenging task in computer vision and graphics. While training-free approaches have been explored, the recent release of the MotionFix dataset, which includes source-text-motion triplets, has opened new avenues for training, yielding promising results. However, existing methods struggle with precise control, often leading to misalignment between motion semantics and language instructions. In this paper, we introduce a related task — motion similarity prediction — and propose a multi-task training paradigm, where we train the model jointly on motion editing and motion similarity prediction to foster the learning of semantically meaningful representations. To complement this task, we design an advanced Diffusion-Transformer-based architecture that separately handles motion similarity prediction and motion editing. Extensive experiments demonstrate the state-of-the-art performance of our approach in both editing alignment and fidelity. Project URL: https://github.com/lzhyu/SimMotionEdit.

Embodied Question Answering (EQA) is an essential yet challenging task for robot assistants. Large vision-language models (VLMs) have shown promise for EQA, but existing approaches either treat it as static video question answering without active exploration or restrict answers to a closed set of choices. These limitations hinder real-world applicability, where a robot must explore efficiently and provide accurate answers in open-vocabulary settings. To overcome these challenges, we introduce EfficientEQA, a novel framework that couples efficient exploration with free-form answer generation. EfficientEQA features three key innovations: (1) Semantic-Value-Weighted Frontier Exploration (SFE) with Verbalized Confidence (VC) from a black-box VLM to prioritize semantically important areas to explore, enabling the agent to gather relevant information faster; (2) a BLIP relevancy-based mechanism to stop adaptively by flagging highly relevant observations as outliers to indicate whether the agent has collected enough information; and (3) a Retrieval-Augmented Generation (RAG) method for the VLM to answer accurately based on pertinent images from the agent’s observation history without relying on predefined choices. Our experimental results show that EfficientEQA achieves over 15% higher answer accuracy and requires over 20% fewer exploration steps than state-of-the-art methods. Our code is available at: https://github.com/chengkaiAcademyCity/EfficientEQA

Obstructive Sleep Apnea (OSA) is a common respiratory disease characterized by recurrent airway block during sleep, which does great harm to the human body. Utilizing the electroencephalogram (EEG) has been proven instrumental in OSA detection, as sleep apnea occurrences induce discernible alterations in EEG patterns. In this study, we propose a Limited Penetrable Difference Visibility Graph (LPDVG) method to screen severe OSA. This method exhibits strong anti-noise performance, effective information extraction, and a certain degree of generalization ability. First of all, this study constructed LPDVG complex network and calculated the information entropy of the degree sequence in six leads. Subsequently, this study weighted the information entropy of each lead using the mutual information between leads to fuse information from the whole brain. Eventually, a classification model for the Graph Convolutional Network (GCN) was trained to detect patients with severe OSA. Using a dataset of 88 participants, we tested and evaluated this approach. The results showed a strong correlation between the extracted feature and AHI, with a Pearson correlation of 0.792. The accuracy, specificity, sensitivity, and area under the curve (AUC) of the GCN classification were 82.95%, 83.87%, 80.77%, and 0.905. Moreover, there are significant differences in LPDVGwSEN between patients in severe and non-severe OSA groups. Compared to those with non-severe OSA, brain activity in patients with severe OSA is more disorganized, especially in the theta frequency band. EEG data indicating this elevated activity is associated with disturbed sleep patterns in individuals with OSA.

This paper addresses a novel task of anticipating 3D human-object interactions (HOIs). Most existing research on HOI synthesis lacks comprehensive whole-body interactions with dynamic objects, e.g., often limited to manipulating small or static objects. Our task is significantly more challenging, as it requires modeling dynamic objects with various shapes, capturing whole-body motion, and ensuring physically valid interactions. To this end, we propose InterDiff, a framework comprising two key steps: (i) interaction diffusion, where we leverage a diffusion model to encode the distribution of future human-object interactions; (ii) interaction correction, where we introduce a physics-informed predictor to correct denoised HOIs in a diffusion step. Our key insight is to inject prior knowledge that the interactions under reference with respect to contact points follow a simple pattern and are easily predictable. Experiments on multiple human-object interaction datasets demonstrate the effectiveness of our method for this task, capable of producing realistic, vivid, and remarkably longterm 3D HOI predictions.