Pub Date : 2021-10-29DOI: 10.1109/ECICE52819.2021.9645737
Yan-Qing Chen, Fu-An Lin, Ting-Yu Yang, S. Yeh, Eric Hsiao-Kuang Wu, J. M. Poole, Charles Shao
Autism Spectrum Disorder (ASD) exhibits social communication and social interaction disorders, and abnormal restrictive and repetitive behaviors. However, symptoms of infants less than 1-year-old are difficult to reliably predict subsequent diagnosis. Patients with mild ASD may not be discovered until school age, because schools have more opportunities for social activities. In addition, the therapist also needs to consider the labor cost. To provide effective treatment, it also needs to consume more resources. The current situation in Taiwan is that outlying islands and remote areas often have insufficient manpower for therapists. If VR technology can be applied, some of the problems may be solved. However, due to the global pandemic, COVID-19, early treatments or group treatments in many countries have been forced to stop. If VR technology can provide interpersonal interaction scenes, the training of ASD children can hardly be affected.This research uses Virtual Reality (VR) technology, combined with wearable multi-model sensing technology, including EEG, eye tracking, heart rate variability (HRV), and breath-sensing strap. Physiological signals and game performance data are collected while users are training, and integrate multiple evaluation scales such as ADOS, SRS, and CBCL. Statistical analysis of these data is performed to classify them through machine learning models to develop a VR assistance system that can be used to evaluate the diagnosis, severity, and social behavior treatment of ASD. This system presents assessment and therapy in a game-oriented way. In addition to enhancing the incentives for users to participate, it provides better training results than traditional training. It is also an effective and convenient tool for the therapist to use during evaluation and training.
{"title":"A VR-based Training and Intelligent Assessment System Integrated with Multi-modal Sensing for Children with Autism Spectrum Disorder","authors":"Yan-Qing Chen, Fu-An Lin, Ting-Yu Yang, S. Yeh, Eric Hsiao-Kuang Wu, J. M. Poole, Charles Shao","doi":"10.1109/ECICE52819.2021.9645737","DOIUrl":"https://doi.org/10.1109/ECICE52819.2021.9645737","url":null,"abstract":"Autism Spectrum Disorder (ASD) exhibits social communication and social interaction disorders, and abnormal restrictive and repetitive behaviors. However, symptoms of infants less than 1-year-old are difficult to reliably predict subsequent diagnosis. Patients with mild ASD may not be discovered until school age, because schools have more opportunities for social activities. In addition, the therapist also needs to consider the labor cost. To provide effective treatment, it also needs to consume more resources. The current situation in Taiwan is that outlying islands and remote areas often have insufficient manpower for therapists. If VR technology can be applied, some of the problems may be solved. However, due to the global pandemic, COVID-19, early treatments or group treatments in many countries have been forced to stop. If VR technology can provide interpersonal interaction scenes, the training of ASD children can hardly be affected.This research uses Virtual Reality (VR) technology, combined with wearable multi-model sensing technology, including EEG, eye tracking, heart rate variability (HRV), and breath-sensing strap. Physiological signals and game performance data are collected while users are training, and integrate multiple evaluation scales such as ADOS, SRS, and CBCL. Statistical analysis of these data is performed to classify them through machine learning models to develop a VR assistance system that can be used to evaluate the diagnosis, severity, and social behavior treatment of ASD. This system presents assessment and therapy in a game-oriented way. In addition to enhancing the incentives for users to participate, it provides better training results than traditional training. It is also an effective and convenient tool for the therapist to use during evaluation and training.","PeriodicalId":176225,"journal":{"name":"2021 IEEE 3rd Eurasia Conference on IOT, Communication and Engineering (ECICE)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131553627","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-29DOI: 10.1109/ECICE52819.2021.9645675
Tangxi Chen, Li-Jung Liu, Yu-Chi Tsao, J. Tsai, T. Wu, Yudan Luo, T. Meen, Chi-Ting Ho
Titanium dioxide (TiO2) is the most commonly used material for the electron transport layer in perovskite solar cells (PSC), but its material defects have affected the development of perovskite solar cells. In this study, cesium carbonate (Cs2CO3) was used to modify the electron transport layer to make it have better electronic conductivity. The structure of the modified perovskite solar cell is FTO/Compact TiO2/mesoporous TiO2/ Cs2CO3/ perovskite layer (MAPbI3)/ sprio-OMETAD/ Au back electrode. The obtained samples were characterized by X-ray diffraction (XRD), ultraviolet/ visible spectrophotometer (UV-Vis), scanning electron microscope (SEM), energy-dispersive X-ray spectroscopy (EDS), and monochromatic incident photon-to-electron conversion. IPCE). In this experiment, six different concentrations of Cs2CO3 were used, 0M (0.90), 0.01M (1.19), 0.02M (1.37), 0.03M (1.55), 0.04M (1.06), 0.05M (0.59). We found 0.03 M cesium is the most suitable concentration for modifying the electron transport layer. Compared to unmodified solar cells, adding a modified layer does not affect the size and thickness of the electron transport layer. The photoelectric conversion efficiency has also increased from 0.90 to 1.55%.
{"title":"Application of Cesium Carbonate Modified Electron Transport Layer to Enhance Performance of Perovskite Solar Cells","authors":"Tangxi Chen, Li-Jung Liu, Yu-Chi Tsao, J. Tsai, T. Wu, Yudan Luo, T. Meen, Chi-Ting Ho","doi":"10.1109/ECICE52819.2021.9645675","DOIUrl":"https://doi.org/10.1109/ECICE52819.2021.9645675","url":null,"abstract":"Titanium dioxide (TiO2) is the most commonly used material for the electron transport layer in perovskite solar cells (PSC), but its material defects have affected the development of perovskite solar cells. In this study, cesium carbonate (Cs2CO3) was used to modify the electron transport layer to make it have better electronic conductivity. The structure of the modified perovskite solar cell is FTO/Compact TiO2/mesoporous TiO2/ Cs2CO3/ perovskite layer (MAPbI3)/ sprio-OMETAD/ Au back electrode. The obtained samples were characterized by X-ray diffraction (XRD), ultraviolet/ visible spectrophotometer (UV-Vis), scanning electron microscope (SEM), energy-dispersive X-ray spectroscopy (EDS), and monochromatic incident photon-to-electron conversion. IPCE). In this experiment, six different concentrations of Cs2CO3 were used, 0M (0.90), 0.01M (1.19), 0.02M (1.37), 0.03M (1.55), 0.04M (1.06), 0.05M (0.59). We found 0.03 M cesium is the most suitable concentration for modifying the electron transport layer. Compared to unmodified solar cells, adding a modified layer does not affect the size and thickness of the electron transport layer. The photoelectric conversion efficiency has also increased from 0.90 to 1.55%.","PeriodicalId":176225,"journal":{"name":"2021 IEEE 3rd Eurasia Conference on IOT, Communication and Engineering (ECICE)","volume":"98 ","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133357007","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this study, a web service and an Android client app that can upload and view live photos on a map are proposed. The study resulted from a collaboration project between the National Yunlin University of Science and Technology and Vietnam National University Ho Chi Minh City - University of Science. The web service provides maps showing many areas. Each map shows different types of positive and negative features. Positive features include buildings, plants, and animals that can help the users to be familiar with the area. When a visitor visits the area, one can use the App to see the pictures of his/her surroundings to identify the location and the correct direction. Negative features include pollutions that users reported to the web service. Examples include chemical and toxic spills, fire, and whether they have been cleaned up or not. Once a pollution site is reported, the appropriate agencies should be notified to check out the condition and clean up the pollution.The success of this web service depends on the active participation of the app users. The quality of the photos and verbal information they upload to the web is critical. The more accurate and precise the information, the better the service will be. Currently, the uploaded information includes the time and the latitude and longitude of the mobile device’s location. Additional functionalities are being considered and designed to improve the service. For example, the user can select the features they want to show on the map and the period when the information was uploaded. Some issues are being considered. For example, how to decide whether the reported object or environmental condition is up-to-date. Specifically, when a pollution site was reported a month ago, how to decide whether the pollution was cleaned up as it is today. The administration can hire an employee to update the information. However, a challenging alternative is to let the app users update the information, and the server can decide whether the information is up-to-date. The web service should be evaluated in future work.
{"title":"Web Service and a Mobile App for Reporting Site Pollution and Other Features","authors":"Winggun Wong, Nguyen Thi Thuy Hang, Meng-Yuan Tsai, Guan-Cheng Shi, Yo-Chen Tsai","doi":"10.1109/ECICE52819.2021.9645628","DOIUrl":"https://doi.org/10.1109/ECICE52819.2021.9645628","url":null,"abstract":"In this study, a web service and an Android client app that can upload and view live photos on a map are proposed. The study resulted from a collaboration project between the National Yunlin University of Science and Technology and Vietnam National University Ho Chi Minh City - University of Science. The web service provides maps showing many areas. Each map shows different types of positive and negative features. Positive features include buildings, plants, and animals that can help the users to be familiar with the area. When a visitor visits the area, one can use the App to see the pictures of his/her surroundings to identify the location and the correct direction. Negative features include pollutions that users reported to the web service. Examples include chemical and toxic spills, fire, and whether they have been cleaned up or not. Once a pollution site is reported, the appropriate agencies should be notified to check out the condition and clean up the pollution.The success of this web service depends on the active participation of the app users. The quality of the photos and verbal information they upload to the web is critical. The more accurate and precise the information, the better the service will be. Currently, the uploaded information includes the time and the latitude and longitude of the mobile device’s location. Additional functionalities are being considered and designed to improve the service. For example, the user can select the features they want to show on the map and the period when the information was uploaded. Some issues are being considered. For example, how to decide whether the reported object or environmental condition is up-to-date. Specifically, when a pollution site was reported a month ago, how to decide whether the pollution was cleaned up as it is today. The administration can hire an employee to update the information. However, a challenging alternative is to let the app users update the information, and the server can decide whether the information is up-to-date. The web service should be evaluated in future work.","PeriodicalId":176225,"journal":{"name":"2021 IEEE 3rd Eurasia Conference on IOT, Communication and Engineering (ECICE)","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132868187","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The existing object detection algorithms have low recognition accuracy for prohibited items due to the complex background, large variation of target scale, and mutual occlusion of objects in X-ray security inspection images. In order to accurately identify prohibited items in real-time, an X-ray security inspection image detection algorithm based on improved YOLOv4 is proposed. Firstly, deformable convolution is introduced into the network to improve the feature extraction ability of prohibited items. Then, GHM loss is used to optimize the loss function, so that the model can focus on the difficult classification samples that are more effective for training improvement. Finally, the non-maximum suppression method combining soft NMS and DIoU NMS is used to improve the detection ability of the algorithm for occluded targets. Experiments on the X-ray security inspection image dataset show that the mAP of the improved algorithm reaches 91.4%, which is 3.3% higher than the YOLOv4, and the detection speed meets the real-time requirements.
{"title":"X-ray Security Inspection Image Detection Algorithm Based on Improved YOLOv4","authors":"Cheng Zhou, Hui Xu, Bicai Yi, Weichao Yu, Chenwei Zhao","doi":"10.1109/ECICE52819.2021.9645636","DOIUrl":"https://doi.org/10.1109/ECICE52819.2021.9645636","url":null,"abstract":"The existing object detection algorithms have low recognition accuracy for prohibited items due to the complex background, large variation of target scale, and mutual occlusion of objects in X-ray security inspection images. In order to accurately identify prohibited items in real-time, an X-ray security inspection image detection algorithm based on improved YOLOv4 is proposed. Firstly, deformable convolution is introduced into the network to improve the feature extraction ability of prohibited items. Then, GHM loss is used to optimize the loss function, so that the model can focus on the difficult classification samples that are more effective for training improvement. Finally, the non-maximum suppression method combining soft NMS and DIoU NMS is used to improve the detection ability of the algorithm for occluded targets. Experiments on the X-ray security inspection image dataset show that the mAP of the improved algorithm reaches 91.4%, which is 3.3% higher than the YOLOv4, and the detection speed meets the real-time requirements.","PeriodicalId":176225,"journal":{"name":"2021 IEEE 3rd Eurasia Conference on IOT, Communication and Engineering (ECICE)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117177803","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-29DOI: 10.1109/ECICE52819.2021.9645620
Shigang Wang, Heng Liu
Due to the design of intravenous imaging instruments and the rapid development of venipuncture robots, it is of great significe to study the high-quality venous image processing method. In this paper, a series of preprocessing of the venous image is carried out, and the development of venous image processing is briefly introduced. The process of vein image processing is mainly divided into three steps. The first step is to use the improved Otsu method for binarization. The second step is to use the selective median filter algorithm to filter. The third step is to refine the image with a fast parallel thinning algorithm. Experimental results show that the vein image processing effect is better.
{"title":"Study of the Preprocessing of Venous Images","authors":"Shigang Wang, Heng Liu","doi":"10.1109/ECICE52819.2021.9645620","DOIUrl":"https://doi.org/10.1109/ECICE52819.2021.9645620","url":null,"abstract":"Due to the design of intravenous imaging instruments and the rapid development of venipuncture robots, it is of great significe to study the high-quality venous image processing method. In this paper, a series of preprocessing of the venous image is carried out, and the development of venous image processing is briefly introduced. The process of vein image processing is mainly divided into three steps. The first step is to use the improved Otsu method for binarization. The second step is to use the selective median filter algorithm to filter. The third step is to refine the image with a fast parallel thinning algorithm. Experimental results show that the vein image processing effect is better.","PeriodicalId":176225,"journal":{"name":"2021 IEEE 3rd Eurasia Conference on IOT, Communication and Engineering (ECICE)","volume":"114 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115806654","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-29DOI: 10.1109/ECICE52819.2021.9645680
Ren-Jwo Tsay
Many old buildings have the problem of weak seismic horizontal load capacity Traditional building retrofit method may take time and must change original structure system. In this paper we developed a steel cable retrofit system which could add external cable system out of original building so the cable will add additional tension force when building take lateral earthquake force. We applied a numerical steel cable reinforcement system in low- and high-rise RC building to understand the different system performance in dynamic loading. From the numerical dynamic analysis of the SAP software shows that the additional steel cable reinforcement around the structure reduces the deformation and structural acceleration in the low- or high-rise building system. So we can applied the results to real building retrofit.
{"title":"Numerical Analysis for the Effectiveness of Buildings Seismic Resistance Capacity by Cable Reinforcement System","authors":"Ren-Jwo Tsay","doi":"10.1109/ECICE52819.2021.9645680","DOIUrl":"https://doi.org/10.1109/ECICE52819.2021.9645680","url":null,"abstract":"Many old buildings have the problem of weak seismic horizontal load capacity Traditional building retrofit method may take time and must change original structure system. In this paper we developed a steel cable retrofit system which could add external cable system out of original building so the cable will add additional tension force when building take lateral earthquake force. We applied a numerical steel cable reinforcement system in low- and high-rise RC building to understand the different system performance in dynamic loading. From the numerical dynamic analysis of the SAP software shows that the additional steel cable reinforcement around the structure reduces the deformation and structural acceleration in the low- or high-rise building system. So we can applied the results to real building retrofit.","PeriodicalId":176225,"journal":{"name":"2021 IEEE 3rd Eurasia Conference on IOT, Communication and Engineering (ECICE)","volume":"55 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123898789","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-29DOI: 10.1109/ECICE52819.2021.9645643
Mingyao Wei, Bo-Rui Lin, Yong-Ye Lin, Gwo-Jen Chiou, W. Kuo
Photoelectric smoke detectors, which detect smoke emitted in the early stages of fire, play a key role in automatic fire alarm equipment. When a fire is detected, the detector transmits an alarm signal to the control panel of the automatic fire alarm equipment. Therefore, the signal intensity and response speed of photoelectric smoke detectors are noteworthy research topics. This study designed an experiment system to explore the effect of different types of smoke on the signal intensity of commercial photoelectric smoke detectors. During a fire, different substances release smoke in different colors while burning. This variation may influence the response speed and sensitivity of smoke detectors. This study conducted an experiment using five types of powders to simulate the smoke release during a fire to test the response of photoelectric smoke detectors. This experimental method is more environmentally friendly as it did not involve any burning substance. The experiment results revealed that the photoelectric smoke detector generated different responses and signals of various intensities when detecting each type of smoke. This finding implies that fire alarm response time differs for different types of smoke. Experiment data revealed that when the detector light source was switched from IR LED to green LED, the sensitivity of the photoelectric smoke detector toward each type of smoke increased. The study findings provide a reference for improving the effectiveness and detection speed of photoelectric smoke detectors.
{"title":"Experimental Study on Effects of Light Source and Different Smoke Characteristics on Signal Intensity of Photoelectric Smoke Detectors","authors":"Mingyao Wei, Bo-Rui Lin, Yong-Ye Lin, Gwo-Jen Chiou, W. Kuo","doi":"10.1109/ECICE52819.2021.9645643","DOIUrl":"https://doi.org/10.1109/ECICE52819.2021.9645643","url":null,"abstract":"Photoelectric smoke detectors, which detect smoke emitted in the early stages of fire, play a key role in automatic fire alarm equipment. When a fire is detected, the detector transmits an alarm signal to the control panel of the automatic fire alarm equipment. Therefore, the signal intensity and response speed of photoelectric smoke detectors are noteworthy research topics. This study designed an experiment system to explore the effect of different types of smoke on the signal intensity of commercial photoelectric smoke detectors. During a fire, different substances release smoke in different colors while burning. This variation may influence the response speed and sensitivity of smoke detectors. This study conducted an experiment using five types of powders to simulate the smoke release during a fire to test the response of photoelectric smoke detectors. This experimental method is more environmentally friendly as it did not involve any burning substance. The experiment results revealed that the photoelectric smoke detector generated different responses and signals of various intensities when detecting each type of smoke. This finding implies that fire alarm response time differs for different types of smoke. Experiment data revealed that when the detector light source was switched from IR LED to green LED, the sensitivity of the photoelectric smoke detector toward each type of smoke increased. The study findings provide a reference for improving the effectiveness and detection speed of photoelectric smoke detectors.","PeriodicalId":176225,"journal":{"name":"2021 IEEE 3rd Eurasia Conference on IOT, Communication and Engineering (ECICE)","volume":"98 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124067613","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-29DOI: 10.1109/ECICE52819.2021.9645629
Po Hsiang Lin, J. Hsieh, Chien-Hua Chen, J. Jeng
Acute chest pain is one of the most common complaints and is frequently related to life-threatening diseases in the emergency department. We aimed to construct a cross-modal deep learning model for risk prediction of acute chest pain by the physicians' clinical texts and electrocardiogram (ECG). Two different modalities included the initial ECG image and the physicians' notes are used to predict the disposition.
{"title":"Cross-Modal Deep Learning Based on Texts and ECG Images for Risk Prediction of Patients with Acute Chest Pain in the Emergency Department","authors":"Po Hsiang Lin, J. Hsieh, Chien-Hua Chen, J. Jeng","doi":"10.1109/ECICE52819.2021.9645629","DOIUrl":"https://doi.org/10.1109/ECICE52819.2021.9645629","url":null,"abstract":"Acute chest pain is one of the most common complaints and is frequently related to life-threatening diseases in the emergency department. We aimed to construct a cross-modal deep learning model for risk prediction of acute chest pain by the physicians' clinical texts and electrocardiogram (ECG). Two different modalities included the initial ECG image and the physicians' notes are used to predict the disposition.","PeriodicalId":176225,"journal":{"name":"2021 IEEE 3rd Eurasia Conference on IOT, Communication and Engineering (ECICE)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116175752","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
To allow UAVs to equip a higher level of autonomous control, this research uses edge computing systems to replace the ground control station commonly used to control UAVs. Since the GCS belongs to the central control architecture, the edge computing system of the distributed architecture gives the drones more flexibility in dealing with changing environmental conditions, allowing them to autonomously and instantly plan their flight path, fly in formation, or even avoid obstacles. Broadcast communications are used to realize UAV-to-UAV communications for allocating tasks among a swarm of UAVs and ensuring each individual collaborates as an integrated member of the group. The dynamic path programming problem for the UAV swarm mission uses a 2-phase Tabu search with the 2-Opt exchange method and A* search as the path programming algorithm. Distance is taken as a cost function for path programming. We then increase and expand the turning-points of no-fly zones based on drone fleet coverage, thus preventing drones from entering prohibited areas. Whereas previous work mostly only considers single no-fly zones, this approach accounts for multiple restricted areas, ensuring that a UAV swarm can complete its assigned task without violating no-fly zones. A drone encountering an obstacle while traveling along the route set by the algorithm will update the map information in real-time, allowing for an instant recharting of the optimal path to the goal as a reverse search using the D* Lite algorithm.
{"title":"UAV Swarm Real-Time Rerouting by Edge Computing under a Changing Environment","authors":"Meng-Tse Lee, Sih-Tse Kuo, Yan-Ru Chen, Ming-Lung Chuang","doi":"10.1109/ECICE52819.2021.9645660","DOIUrl":"https://doi.org/10.1109/ECICE52819.2021.9645660","url":null,"abstract":"To allow UAVs to equip a higher level of autonomous control, this research uses edge computing systems to replace the ground control station commonly used to control UAVs. Since the GCS belongs to the central control architecture, the edge computing system of the distributed architecture gives the drones more flexibility in dealing with changing environmental conditions, allowing them to autonomously and instantly plan their flight path, fly in formation, or even avoid obstacles. Broadcast communications are used to realize UAV-to-UAV communications for allocating tasks among a swarm of UAVs and ensuring each individual collaborates as an integrated member of the group. The dynamic path programming problem for the UAV swarm mission uses a 2-phase Tabu search with the 2-Opt exchange method and A* search as the path programming algorithm. Distance is taken as a cost function for path programming. We then increase and expand the turning-points of no-fly zones based on drone fleet coverage, thus preventing drones from entering prohibited areas. Whereas previous work mostly only considers single no-fly zones, this approach accounts for multiple restricted areas, ensuring that a UAV swarm can complete its assigned task without violating no-fly zones. A drone encountering an obstacle while traveling along the route set by the algorithm will update the map information in real-time, allowing for an instant recharting of the optimal path to the goal as a reverse search using the D* Lite algorithm.","PeriodicalId":176225,"journal":{"name":"2021 IEEE 3rd Eurasia Conference on IOT, Communication and Engineering (ECICE)","volume":"37 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115081913","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-10-29DOI: 10.1109/ECICE52819.2021.9645706
Guoling Cui
Deep convolutional neural network is one of the most popular research topics in the field of computer vision. It has the function of extracting image feature information, has strong nonlinear classification ability, fast learning speed, and can be used for image recognition and classification. This paper makes use of its image recognition and classification function to carry on the research of its recognition and classification technology in oil painting schools. Through the ResNet network structure of a deep convolutional neural network, a data set is constructed by load data function, and then embedded into a SEBlock model, the accuracy and generalization ability of image recognition and classification of the deep convolutional neural network can be greatly improved. Among them, the SE model has strong effectiveness and generalization ability. For example, the accuracy of the SE-ResNet-34 is 1.73% higher than that of the ResNet-34, and the accuracy of the SE-ResNet-50 has reached that of the ResNet-101. The SE model is applied to the deep convolutional neural network to improve classification accuracy and reduce errors.
{"title":"Research on Recognition and Classification Technology Based on Deep Convolutional Neural Network","authors":"Guoling Cui","doi":"10.1109/ECICE52819.2021.9645706","DOIUrl":"https://doi.org/10.1109/ECICE52819.2021.9645706","url":null,"abstract":"Deep convolutional neural network is one of the most popular research topics in the field of computer vision. It has the function of extracting image feature information, has strong nonlinear classification ability, fast learning speed, and can be used for image recognition and classification. This paper makes use of its image recognition and classification function to carry on the research of its recognition and classification technology in oil painting schools. Through the ResNet network structure of a deep convolutional neural network, a data set is constructed by load data function, and then embedded into a SEBlock model, the accuracy and generalization ability of image recognition and classification of the deep convolutional neural network can be greatly improved. Among them, the SE model has strong effectiveness and generalization ability. For example, the accuracy of the SE-ResNet-34 is 1.73% higher than that of the ResNet-34, and the accuracy of the SE-ResNet-50 has reached that of the ResNet-101. The SE model is applied to the deep convolutional neural network to improve classification accuracy and reduce errors.","PeriodicalId":176225,"journal":{"name":"2021 IEEE 3rd Eurasia Conference on IOT, Communication and Engineering (ECICE)","volume":"82 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117057504","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：