Yin-Tien Wang

For small robots, annotation can outlast computation as a bottleneck: hardware is bought once, labels at every deployment. We study label efficiency for landmark detection on a humanoid soccer robot. Against soft-argmax and pooled coordinate-regression baselines on 10–100% of 1193 labeled images under one protocol and two held-out sessions, the heatmap configuration is the more label-efficient: with hyperparameters fixed on the full pool, 298 training images (509 annotated) give 0.436 PCK@10 px, which neither baseline reaches with all 1193. The heatmap and soft-argmax arms differ in target, loss and readout, so the interval compares configurations, not representations; at equal held-out false-detection rates the heatmap stays ahead at both tolerances. A validation-to-held-out gap widens from 0.245 to 0.350 PCK@10 with more labels, similarly on both sessions. Int8 conversion, on the host CPU for two checkpoints, changes PCK@10 by at most 0.003. Two Edge TPU accelerators run the detectors in 5.18 and 19.71 ms, timed alone; reduced host bandwidth slows them by 2.8 and 8.2 for the landmark network despite near-complete parameter caching. A Monte Carlo filter is accurate to centimeters in simulation; one field demonstration agrees with hand-annotated positions only to decimeters, with an unresolved frame offset. Loop latency is unmeasured.

6D object pose estimation is an essential component for robotic grasping. Most existing deep learning-based approaches focus on instance-level pose estimation, which requires prior object models and consequently limits their applicability on unseen objects in real-world scenarios. In contrast, category-level 6D pose estimation adopts Normalized Object Coordinate Space (NOCS) maps to represent intra-class object geometry, enabling pose prediction without relying on predefined object models and thus improving generalization to unseen instances. However, the original NOCS-based category-level framework typically trains NOCS prediction and object classification in a joint manner, which introduces NOCS regression error among inter-class instances with similar appearances, thereby degrading pose estimation accuracy. To address this issue, we integrate the YOLOv8 object detection with SegFormer and propose a novel Category-Level SegFormer for 6D Object Pose Estimation (CLSF-6DPE). By decoupling object classification from NOCS regression through independent learning branches, the proposed framework significantly improves pose estimation performance. Furthermore, we validate the practical feasibility of CLSF-6DPE by integrating it with a robotic gripper via the Robot Operating System (ROS) in a Real-World grasping setup. Experimental results on the CAMERA and Real-World datasets demonstrate that the proposed method achieves mAP scores of 93.8% and 81.1%, respectively. Overall, the proposed method provides a modular and effective solution for category-level pose estimation in real-world robotic grasping applications.

Defect detection in mass production lines often involves small and imbalanced datasets, necessitating the use of few-shot learning methods. Traditional deep learning-based approaches typically rely on large datasets, limiting their applicability in real-world scenarios. This study explores few-shot learning models for detecting product defects using limited data, enhancing model generalization and stability. Unlike previous deep learning models that require extensive datasets, our approach effectively performs defect detection with minimal data. We propose a Siamese network that integrates Residual blocks, Squeeze and Excitation blocks, and Convolution Block Attention Modules (ResNet-SE-CBAM Siamese network) for feature extraction, optimized through triplet loss for embedding learning. The ResNet-SE-CBAM Siamese network incorporates two primary features: attention mechanisms and metric learning. The recently developed attention mechanisms enhance the convolutional neural network operations and significantly improve feature extraction performance. Meanwhile, metric learning allows for the addition or removal of feature classes without the need to retrain the model, improving its applicability in industrial production lines with limited defect samples. To further improve training efficiency with imbalanced datasets, we introduce a sample selection method based on the Structural Similarity Index Measure (SSIM). Additionally, a high defect rate training strategy is utilized to reduce the False Negative Rate (FNR) and ensure no missed defect detections. At the classification stage, a K-Nearest Neighbor (KNN) classifier is employed to mitigate overfitting risks and enhance stability in few-shot conditions. The experimental results demonstrate that with a good-to-defect ratio of 20:40, the proposed system achieves a classification accuracy of 94% and an FNR of 2%. Furthermore, when the number of defective samples increases to 80, the system achieves zero false negatives (FNR = 0%). The proposed metric learning approach outperforms traditional deep learning models, such as parametric-based YOLO series models in defect detection, achieving higher accuracy and lower miss rates, highlighting its potential for high-reliability industrial deployment.

Efficient and accurate license plate recognition (LPR) in unconstrained environments remains a critical challenge, particularly when confronted with skewed imaging angles and the limited computational capabilities of edge devices. In this study, we propose a high-performance, FPGA-based license plate alignment and recognition (LPAR) system to address these issues. Our LPAR system integrates lightweight deep learning models, including YOLOv4-tiny for license plate detection, a refined convolutional pose machine (CPM) for pose estimation and alignment, and a modified LPRNet for character recognition. By restructuring the pose estimation and alignment architectures to enhance the geometric correction of license plates and adding channel and spatial attention mechanisms to LPRNet for better character recognition, the proposed LPAR system improves recognition accuracy from 88.33% to 95.00%. The complete pipeline achieved a processing speed of 2.00 frames per second (FPS) on a resource-constrained FPGA platform, demonstrating its practical viability for real-time deployment in edge computing scenarios.

Recent advancements in additive manufacturing have been propelled by innovations in processes, materials, and fabrication techniques. This study specifically focuses on the optimizing Wire and Arc Additive Manufacturing (WAAM) within the Gas Metal Arc Welding (GMAW) process. The goal is to build laminated layers using aluminum alloy from the five-thousand series. In this research, a novel technique is introduced that employs a reduced arc-current during the additive layer progress for aluminum alloys, along with a double pulse current in the GMAW process to enhance weldability. The material used for addition is ER5356 aluminum wire, which is applied to an Al-5083 substrate. The study successfully optimized a close-path structure comprising up to 20 layers. These layers serve as the basis for evaluating laminate formation, effective dimension, and imperfection occurrence to achieve a near-net shape with low additive defects. A thorough evaluation was conducted to identify the suitable conditions for successful laminate build-up. Additionally, a metallography analysis was performed to examine the chemical distribution and microstructure grain size. The results revealed insignificant chemical distribution and composition along the cross-section of the additive layer under suitable conditions. Furthermore, the grain size measurements indicate that lower reduced arc current conditions led to the formation of finer grains, implying superior mechanical properties. This research provides significant insights into the enhancement of GMAW for aluminum additive manufacturing, underscoring the importance of process parameters, material choice, and fabrication techniques in obtaining improved results in laminate formation, microstructure, and mechanical performance. The developed WAAM laminated layers are particularly suitable for fabricating or repairing large-scale components such as ship propellers, hull sections, and offshore platform structures. Additionally, the corrosion-resistant properties of aluminum alloys make them suitable for applications in marine and offshore environments.

Most online action detection methods focus on solving a (K + 1) classification problem, where the additional category represents the ‘background’ class. However, training on the ‘background’ class and managing data imbalance are common challenges in online action detection. To address these issues, we propose a framework for online action detection by incorporating an additional pathway between the feature extractor and online action detection model. Specifically, we present one configuration that retains feature distinctions for fusion with the final decision from the Long Short-Term Transformer (LSTR), enhancing its performance in the (K + 1) classification. Experimental results show that the proposed method achieves an accuracy of 71.2% in mean Average Precision (mAP) on the Thumos14 dataset, outperforming the 69.5% achieved by the original LSTR method.

As the Industry 4.0 landscape unfolds, collaborative robots (cobots) play an important role in intelligent manufacturing. Compared with industrial robots, cobots are more flexible and intuitive in programming, especially for industrial and home service applications; However, there are still issues to be solved, including understanding human intention in a natural way, adaptability to execute tasks, and robot mobility in a working environment. As an attempt to solve the problems aforementioned, in this paper, we propose a modularized solution for mobile cobot systems, where the cobot equipped with a multi-camera localization scheme for self-localization can understand the human intention via human voice commands to execute the tasks in an unseen scenario in a small-area working environment. As far as intention understanding is concerned, we devise a natural language processing approach to establish an action base to describe human commands. According to the action base, the robot can then execute the tasks by planning a trajectory with the help of an object localization module, which integrates the point cloud and the object detected by YOLOv4 to locate the object’s position in three-dimension space. Depending on where the cobot interacts with the object, the cobot might need to navigate around the working environment. Thus, we also establish a low-cost and high-efficiency multi-camera localization system with ArUco markers to locate the mobile cobot in a larger sensing area. Experimental results show that the proposed vision-based mobile cobot can successfully interact with a human operator to assemble a wooden chair in a small workshop.

Stainless steel wires with fiber-laser welding on lap joints are first proposed in this study to be used as cerclage wire joints for modern femur surgery, because of their potentially larger joint strength and less loosening failure than traditional joint devices. In this feasibility study, an experiment was set up to determine adequate parameters for the laser welding process in order to ensure that the wire joint of cerclage has good weld appearance, free oxidation, and suitable joint strength. A stainless steel wire 316L with a diameter of 1.6 mm, flare-welded on lap joint was used in the experiment as a specimen cerclage wire joint. Two major effective parameters were chosen for controlling a suitable fusion weld, including charge voltage and multiple frequencies of the laser irradiation. The adequate area of the laser parameters was determined from the experiment, including the ranges of charge voltage, multiple frequencies, and pulse width. The suitable welded lengths of specimens were also studied in the mechanical test to validate the joint strength. Suggested welded length has a better tensile strength than traditional cerclage joints. The paper concludes that the stainless wire joints with a fiber laser weld represents a promising alternative to traditional cerclage joint devices for modern femur surgery.