http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Wood Polish Classification for Automated Quality Inspection based on AI Vision
Hsien-I Lin,Satrio Dwi Sanjaya 제어로봇시스템학회 2021 제어로봇시스템학회 국제학술대회 논문집 Vol.2021 No.10
Nowadays, the demand for quality inspection of wood polishing is increasing. Thus, there is a need on industrial level to maintain high quality inspection. The quality inspection on wood polishing is currently done by human labors, which is inefficient, costly, and time-consuming. To reduce the cost of wood quality inspection, we propose an automated quality inspection based on AI vision to distinguish whether the wood is polished or unpolished. This system uses a deep learning method to classify polished or unpolished wood, which is one of the pioneer works using deep learning to examine wood quality. In this paper, we adopt the Efficient Net architecture because of its superior capability of handling the model parameters. The proposed approach combines Adam optimizer and SoftMax classifiers to provide the better performance of the model. This paper presents the binary classification on our dataset that contains 1,920 training and 560 test images. The result showed an average accuracy of 85%. In addition, the Efficient Net indicated the competitive performance metric of 85% as recall, 85.5% as precision, and 85% as f1-score. In conclusion, the proposed architecture is satisfactory for automated quality inspection in the wood polishing process.
Learning Human-Robot Collaboration with POMDP
Hsien-I Lin,Xuan-Anh Nguyen 제어로봇시스템학회 2016 제어로봇시스템학회 국제학술대회 논문집 Vol.2016 No.10
Modeling and planing a collaborative task for a human and a robot in an unstructured environment is a challenging problem. Most of earlier works simplify the task by assuming exact knowledge of the environment and the human intention, little work has been done on how to plan a collaborative task in the presence of uncertainty. However, there still exists a lot of challenges because of limited perception abilities. To reason explicitly about uncertainty during collaboration between a human and a robot in a household task, we here present a decision making approach for human robot interactivity based on a Partially Observable Markov Decision Process (POMDP). The approach is validated in simulation in a collaborative scenario of assisting a person make a cup of coffee.
Inference of 6-DOF Robot Grasps using Point Cloud Data
Hsien-I Lin,Minh Nguyen Cong 제어로봇시스템학회 2019 제어로봇시스템학회 국제학술대회 논문집 Vol.2019 No.10
Recent decade, grasping pose perception for robot grasping field takes an important role in developing intelligent robotic applications. In this paper, we present a novel pipeline to infer grasp configuration for 3-finger gripper from point cloud data of a working environment. Most of the previous works are on 2D or 2.5D inputs to perform the 2D grasps. However, the 2D grasp configuration is not really compatible with the physical robot structure, the grasp successful rate on a robot system is only around 73% to 80%. To improve the robustness and computation, our goal is to propose an approach that can generate reliable grasping configurations in the cluster space. To accomplish this, we adopt a point cloud deep network to learn the shapes of objects and recognize their surfaces from the partial view. Afterwards, the grasp candidates are inferred based on the object models and reference grasping configuration sets. In our experiment, the proposed point cloud network had a better successful grasping rate compared to previous approaches in the context of varied and imbalanced data. And the grasping inference was highly viable in the complicated working environment.
Semantic Recognition of Human Gestures Based on Spatial and Temporal Reasoning
Hsien-I Lin,Wei-Kai Chen,C. N. Huang,H. W. Wang 제어로봇시스템학회 2013 제어로봇시스템학회 국제학술대회 논문집 Vol.2013 No.10
Gesture semantic recognition is a vital process of motion recognition. An accurate and stable recognition method helps to identify semantic meaning of a continuous motion. In this work, the semantic meaning is represented by a sequence of motion primitives. This work proposes a method based on spatial and temporal reasoning. Both spatial and temporal rules of motion primitives are used to estimate the current state of motion. The motion primitives adopted in this work are transport empty (TE), transport loaded (TL), grasp (G), and release (RL) from Gilbreth’s therbligs. The results show that the average recognition rate is 94.44% out of three different tasks.
Robotic Arm Path Planning Based on Three-Dimensional Artificial Potential Field
Hsien-I Lin,Ming-Feng Hsieh 제어로봇시스템학회 2018 제어로봇시스템학회 국제학술대회 논문집 Vol.2018 No.10
This paper focuses on the problem of collision avoidance by an three-dimensional artificial potential field (APF) for a robotic arm. In this paper, a force sensor is adopted to estimate obstacle positions on a reference trajectory to a target. This helps a robotic arm move in a safe path in an unknown 3D environment. Compared to traditional APF methods, the major contribution of our proposed algorithm is the implementation of rotational repulsive force fields that provide a robotic arm a feasible direction to avoid obstacles in a 3D space. The experimental results validated that the robotic arm could avoid unknown obstacles on the reference trajectory until the target was reached. Also, the robotic arm learned to improve the trajectory according to the previous trials in a short-term run.
Behavior-Based Manipulator Programming Based on Extensible Agent Behavior Specification Language
Hsien-I Lin,Chia-Hsien Cheng 제어로봇시스템학회 2014 제어로봇시스템학회 국제학술대회 논문집 Vol.2014 No.10
Endowing a robot with skills to perform manipulative tasks has an important role in developing an intelligent robot. To manipulate objects, a robot needs perception and action skills. However, designing the programming framework to integrate a variety of skills in a robot system is a challenging task and significantly influences the robot performance. In this paper, we present a behavior-based manipulator programming framework which is based on Extensible Agent Behavior Specification Language (XABSL) to manage behaviors in a robot system. To achieve the flexibility and reusability of robot behaviors required for practice applications, the proposed concept is to implement a programming framework for robot manipulation into two steps: first, perception and action behaviors are created to endow a robot with fundamental skills to perform manipulative tasks; second, using the XABSL framework, the created behaviors are simply planned by an option graph. Because behaviors are planned to be activated by certain stimuli and respond accordingly, programming robot manipulative tasks becomes simpler. Moreover, by the programming framework for robot manipulative tasks, the programming effort is reduced considerably. In our experiments, we provide an extensive validation of the proposed behavior-based programming framework on the manipulative tasks such as stacking cubes and solving rubik’s cube.
Shih-Hsien Lin,Huai-Hsuan Tseng,Hsin Chun Tsai,Mei Hung Chi,I Hui Lee,Po See Chen,Kao Chin Chen,Yen Kuang Yang 대한정신약물학회 2021 CLINICAL PSYCHOPHARMACOLOGY AND NEUROSCIENCE Vol.19 No.1
Objective: Weight gain is an important risk factor for morbidity and mortality among patients with schizophrenia. We speculated that positive symptoms, related to dopaminergic hyperactivity and altered mesolimbic function, are associated with weight gain. Methods: Twenty-two antipsychotic-naïve, first-episode patients with schizophrenia were enrolled. The Positive and Negative Syndrome Scale was completed at enrollment and follow-up. Body mass index (BMI) was also measured. Results: The increase in BMI, after 6.04 ± 2.16 years of follow-up, was associated with positive symptoms, but not negative symptoms, before treatment with antipsychotics in antipsychotic-naïve patients with schizophrenia. Conclusion: This finding implied that dopaminergic hyperactivity could be an important factor to predict the treatment outcome. Body weight control is important for the health management of patients with schizophrenia with more severe positive symptoms.
Ting-I Chiang,Yi-Hsiang Yu,Chieh-Hsin Lin,Hsien-Yuan Lane 대한정신약물학회 2021 CLINICAL PSYCHOPHARMACOLOGY AND NEUROSCIENCE Vol.19 No.3
Early detection and prevention of Alzheimer’s disease (AD) is important. The current treatment for early AD is acetylcholine esterase inhibitors (AChEIs); however, the efficacy is poor. Besides, AChEI did not show efficacy in mild cognitive impairment (MCI). Beta-amyloid (A) deposits have been regarded to be highly related to the pathogenesis of AD. However, many clinical trials aiming at the clearance of A deposits failed to improve the cognitive decline of AD, even at its early phase. There should be other important mechanisms unproven in the course of AD and MCI. Feasible biomarkers for the diagnosis and treatment response of AD are lacking to date. The N-methyl-D-aspartate receptor (NMDAR) activation plays an important role in learning and memory. On the other hand, oxidative stress has been regarded to contribute to aging with the assumption that free radicals damage cell constituents and connective tissues. Our recent study found that an NMDAR enhancer, sodium benzoate (the pivotal inhibitor of D-amino acid oxidase [DAAO]), improved the cognitive and global function of patients with early-phase AD. Further, we found that peripheral DAAO levels were higher in patients with MCI and AD than healthy controls. We also found that sodium benzoate was able to change the activity of antioxidant. These pieces of evidence suggest that the NMDAR function is associated with anti-oxidation, and have potential to be biomarkers for the diagnosis and treatment response of AD.
The back contact modification in high-efficiency Cu₂ZnSn(S,Se)₄ solar cells by a thin MoO₃ layer
Septia KHOLIMATUSSADIAH,Cheng-Ying CHEN,Wei-Chao CHEN,Yi-Rung LIN,Shao-Hung LU,Meng-Chia HSIEH,Jan-Kai CHANG,Chih-I WU,Ruei-San CHEN,Kuei-Hsien CHEN,Li-Chyong CHEN 한국진공학회 2016 한국진공학회 학술발표회초록집 Vol.2016 No.8