Ai-Hui Wang

Information-Preserving Lightweight Object Detection Network Utilizing Dynamic Gating and Multi-Scale Selection

Visual object detection is essential for environment perception in intelligent robots, automated assembly, unmanned inspection, and industrial detection systems. Although lightweight detectors reduce complexity through compact architectures, fixed convolutional units, and progressive downsampling, their limited scale responses and spatial-detail loss may degrade the localization of scale-varying and boundary-sensitive objects. To address this issue, this paper proposes DGMS-YOLO, an information-preserving lightweight detector built on a one-stage framework. The model improves feature representation through adaptive scale selection and information preservation. Specifically, the Dynamic Gated Multi-Scale Selection (DGMS) module extracts multi-scale features using depthwise convolution branches with different receptive fields and generates content-aware scale weights from the mean and standard deviation of input features. A temperature-scaled softmax and uniform scale prior are further introduced to prevent premature branch-weight concentration during multi-branch training. The Dual Pooling Downsampling (DPD) module combines max pooling, average pooling, and stride convolution to preserve salient responses, regional structures, and learnable semantic features during downsampling. In addition, high-frequency residual calibration estimates edge residuals from low-frequency features and applies lightweight channel gating to compensate for localization-related textures and boundary details. On PASCAL VOC 2007, DGMS-YOLO achieves $\text{7 6. 6 7 \%} \text{m A P}_{50}$ and $\text{5 5. 3 3 \%} \text{m A P}_{50: 95}$ with a parameter count comparable to YOLOv8s, improving it by 1.23 and 2.00 percentage points, respectively. These results demonstrate that dynamic scale selection and information preservation improve detection accuracy and localization quality under a parameter and storage budget comparable to YOLOv8s.

Error Compensation Strategies for Lower-Limb Rehabilitation Robots: A Staged Approach with MLP and Transformer Models

This study significantly reduces the gait trajectory tracking errors of joint actuators in a lower-limb rehabilitation robot, thereby providing a feasible and effective approach for the optimization of its control algorithm design.

EEG-Based Brain-Computer Interface Control for Lower-Limb Rehabilitation Robots: A Focused Review

Electroencephalography-based brain-computer interfaces (EEG-based BCIs) provide a non-invasive pathway for incorporating voluntary neural activity into lower-limb rehabilitation robots, exoskeletons, robotic orthoses, and gaittraining systems. This focused review examines the MI-based brain-robot rehabilitation loop, including lower-limb intention decoding, high-level command generation, robot or gait-device interaction, and closed-loop feedback. Evidence is interpreted across four categories: offline lower-limb MI decoding, online BCI demonstrations, lower-limb device integration, and patient-oriented or clinical evaluation. Representative studies support the technical feasibility of decoding lower-limb motor imagery (MI) and using selected BCI outputs in virtual-reality, exoskeleton, and treadmill systems. However, the evidence remains dominated by offline analyses and small proof-of-concept studies, with limited standardized clinical outcomes. Hybrid sensing and multimodal feedback have been explored as complementary strategies, but their value in physical lower-limb rehabilitation requires direct online and patient-oriented validation. The review therefore distinguishes transferable decoding advances from direct rehabilitation evidence and identifies priorities for safe and clinically meaningful system development.

Adaptive SMC for Lower Limb Rehabilitation Robots Using a Sliding Mode Hyperbolic ESO