Indoor positioning is crucial for autonomous robots, especially in indoor human-robot collaborative laboratories where moderate personnel activity, portable experimental equipment, illumination variation, and local weak-texture regions coexist. However, accurate indoor positioning remains challenging because the robot must maintain reliable perception under dynamic interference, non-structural clutter, and appearance changes. This paper presents a two-stage matching-based visual relocalization framework for indoor mobile robots operating in pre-mapped environments. First, a multi-scale variational autoencoder latent representation model is developed to evaluate scene similarity between images, enabling the robot to recognize the current scene from a pre-constructed image-based map and thereby achieve coarse localization. Then, a SuperGlue-based keypoint matching approach is utilized to calculate the relative pose between the current image and a reference image with known global pose, yielding precise global localization. Experimental results in three representative challenging scenes demonstrate that the proposed framework achieves a coarse localization accuracy of 99.00%. For precise localization, it attains mean absolute translation errors of 0.076 m, 0.065 m, and 0.033 m along the X, Y, and Z axes, respectively, as well as average rotation errors of 0.66°, 0.70°, and 0.73° in roll, pitch, and yaw, respectively.

The surge of data center loads and high renewable penetration threatens power system stability, requiring joint power-computing scheduling. This paper presents an attention-driven hierarchical multi-agent proximal policy optimization (H-MAPPO) method to solve the coordination issue. A Transformer-based fine-grained task-computing-power coupling model is built to map diverse computing tasks to grid loads. The grid-computing interaction is formulated as a Stackelberg game via bi-level Markov decision process (MDP). The upper agent conducts global dispatch with proximal policy optimization (PPO), and lower agents use attention-based coordination algorithm PPO (ACA-PPO) for distributed task scheduling. Tests on German European network of transmission system operators for electricity (ENTSO-E) grid data and Google cluster traces reveal H-MAPPO attains 98.36% profit of the optimal mixed integer quadratic programming (MIQP), shortens solving time from 342.6 s to 2.5 s, and lifts task completion rate by 21.6% versus hierarchical independent proximal policy optimization (H-IPPO). The hierarchical attention-aided framework supports scalable real-time power-computing collaborative scheduling for low-carbon grids.

This study addresses the issue of spatial coverage blind spots induced by the limited three-dimensional (3D) shear wave (S-wave) velocity resolution of conventional surface wave imaging methods. To resolve this problem, a 3D S-wave velocity forecasting method based on large language models (LLMs) is proposed. Specifically, we design a prompting template tailored to the 3D S-wave velocity forecasting task combined with a 3D S-wave velocity field grid traversal strategy. The approach converts previously inverted 3D S-wave velocity data into a 3D S-wave velocity prompt sequence dataset with spatiotemporal sequence characteristics. Then, by fine-tuning the LLM using a standard trainer, the fine-tuned LLM is leveraged to forecast S-wave velocities in areas of the study region where coverage is incomplete. Experimental results demonstrate that LLM models ( after fine-tuning via prompting techniques and optimization through data engineering) exhibit a good forecasting performance and generalization ability in the scenario of 3D S-wave velocity prediction. These results validate the feasibility and reliability of the proposed method. The findings of this study can be applied to high-resolution S-wave velocity imaging and provide a valuable reference for in-depth investigations of complex geological structures within the geoscience field.

Adversarial attacks are a major challenge to the security of deep neural networks (DNNs). However, current transferable attacks often produce perturbations deterministically. This leads to poor local optima, resulting in low transferability. To address this issue, an attention-space transfer attack (ASTA) is proposed to enhance transferability through randomized disruption of salient features. Specifically, a randomized perturbation strategy utilizing latent codes is employed to disrupt shared attention features across models. Furthermore, attention and diversity loss functions are introduced to explicitly guide the generator in exploring the perturbation search space, thereby avoiding overfitting to surrogate models. Experimental results demonstrate that this method achieves an average attack success rate (ASR) of 83.6% on multiple black-box target ensemble models, 6.7% higher than the baseline methods. It effectively generates adversarial samples and demonstrates stronger transferability.

To ensure the confidentiality of Internet of things (IoT) cloud data while enabling fine-grained access control, attribute-based encryption (ABE) schemes have been widely adopted. Despite their advantages, existing approaches suffer from three critical shortcomings: reliance on a central authority (CA) for key distribution which may lead to a single point of failure, dependence on mathematical hard problems like discrete logarithms and bilinear pairings which leaves most of them vulnerable to quantum attacks, and insufficient attention to user privacy protection mechanisms in current lattice-based ABE designs. To address these limitations, this article proposes a privacy-preserving personal health record (PHR) sharing framework based on lattice-based multi-authority attribute-based encryption (MA-ABE). The framework enhances user privacy through an anonymous commitment protocol that conceals identities and a key generation algorithm that allows decentralized authorities to issue private keys without exposing user information. Furthermore, the design extends beyond conventional single-attribute-per-authority models, enabling each authority to manage multiple attributes. Moreover, two-way anonymity for both provers and verifiers is achieved through these user privacy protection mechanisms. The proposed scheme achieves indistinguishability under selective-attribute chosen plaintext attack (IND-sAtt-CPA) security, while maintaining acceptable computational efficiency.

Task scheduling of directed acyclic graph (DAG) based applications in edge computing faces challenges from dynamic task variations and lack of deterministic latency guarantees, while traditional algorithms perform poorly in latency, energy consumption and resource utilization, making them unable to meet the requirements of latency-sensitive applications. To address this issue, this paper proposes the network calculus soft actor-critic (NCSAC) algorithm, which integrates network calculus with the soft actor-critic (SAC) algorithm for DAG task scheduling. NCSAC formulates the DAG scheduling problem as a Markov decision process (MDP) and constructs a joint model of task dependency graphs and edge node networks, using dual graph convolutional networks (DGCNs) to extract deep features. It embeds network calculus into the policy network loss function with a delay penalty to prioritize latency-constrained scheduling policies, and employs SAC's maximum entropy framework, soft update and experience replay to balance exploration and exploitation for stable training. A multi-objective optimization objective is designed to optimize delay deviation and network traffic cost simultaneously. Comparative experiments are conducted with network-calculus-based proximal policy optimization (NCPPO), SAC, min-min scheduling algorithm and heterogeneous earliest finish time (HEFT) algorithm. Results demonstrate that NCSAC achieves low and steadily increasing total latency deviation and worst-case latency under dynamic DAG tasks and edge nodes, reduces network cost by 15.4% - 48.0% on average, and exhibits superior latency determinism, cost efficiency and robustness.

This study addresses hybrid traffic scheduling challenges in time-sensitive networking (TSN) over wide area networks (WANs) by proposing a dynamic collaborative scheduling algorithm based on cyclic specified queuing and forwarding (CSQF). Fixed resource blocks (FBs) and elastic resource blocks (EBs) are partitioned to isolate time-triggered (TT) and audio-video bridging (AVB) flows. Dynamic cycle scaling and sliding window preemption further optimize resource efficiency. Simulations show the algorithm achieves 75.6% AVB flow scheduling success under 3 000 total flows, reduces resource fragmentation by 28% compared to static strategies, and ensures 100% TT flow success. Results demonstrate its effectiveness in balancing strict TT delay constraints with enhanced hybrid traffic performance, offering a practical solution for deterministic WAN transmission.

With the aging population, demand for artificial joint replacement is growing. In total hip arthroplasty (THA), implanting the acetabular cup into the safe zone (inclination: 40° ± 10°, anteversion: 15° ± 10°) is critical for success, but existing technologies suffer from low accuracy or radiation invasiveness. To address this, a non-invasive prosthesis pose visualization method based on multi-sensor fusion is proposed, aiming for real-time, radiation-free and high-precision monitoring. A complete hardware-software system is constructed, consisting of an inertial measurement unit (IMU), a microprocessor, a bluetooth transceiver and a computer. An extended Kalman filter (EKF) is employed for IMU data fusion, and a non-invasive registration method is proposed to establish the coordinate transformation, enabling calculation and visualization of the cup's inclination and anteversion. Experiments showed that the root mean square error (RMSE) of self-pose estimation was within 0.423°, the RMSE of the inclination and anteversion were 0.278° and 0.296° respectively, and the average error vector magnitude was 0.373°, all within the safe zone tolerance, rendering the proposed method promising as an effective surgical tool.