Pub Date : 2017-11-01DOI: 10.1109/MFI.2017.8170413
Abdelhameed A. A. Zayed, Samy F. M. Assal, A. F. El-Bab
Harvesting energy from mechanical vibrations to provide power in remote areas where there is no lines of electricity has attracted a great research interest. Linear vibration energy harvesters (VEH) are not able to scavenge energy over broadband of frequencies. The performance of the linear harvesters can be improved using the concept of nonlinearity using magnets which has been extensively studied. In this paper, a 2-DOF nonlinear VEH that can maximize the harvested ambient energy over a wide bandwidth is proposed. In order to achieve those requirements, the design of the proposed VEH is based on; first, the cut-out structure of the 2-DOF system that can provide two resonances with significantly high amplitudes; second, parameters selection based on satisfying the dynamic vibration absorber condition that can avoid the anti-resonance between the two peaks while maximizing the response in between those two peaks; third, proper selection of the distance between the two magnets to widen the bandwidth through adding the nonlinearity to the system. Simulations for the proposed harvester are carried out using MATLAB/Simulink over a wide bandwidth of frequencies. The results show that the proposed harvester can generate adequate voltage of 5 volts across a load of 1 GΩ over a wide range of frequency from 26 to 35 Hz for the proper distance between the two magnets that is based on the system parameters. Also, another frequency range from 5 to 11 Hz is targeted to insure that the proposed design methodology can be generalized.
{"title":"Wide bandwidth nonlinear 2-DOF energy harvester: Modeling and parameters selection","authors":"Abdelhameed A. A. Zayed, Samy F. M. Assal, A. F. El-Bab","doi":"10.1109/MFI.2017.8170413","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170413","url":null,"abstract":"Harvesting energy from mechanical vibrations to provide power in remote areas where there is no lines of electricity has attracted a great research interest. Linear vibration energy harvesters (VEH) are not able to scavenge energy over broadband of frequencies. The performance of the linear harvesters can be improved using the concept of nonlinearity using magnets which has been extensively studied. In this paper, a 2-DOF nonlinear VEH that can maximize the harvested ambient energy over a wide bandwidth is proposed. In order to achieve those requirements, the design of the proposed VEH is based on; first, the cut-out structure of the 2-DOF system that can provide two resonances with significantly high amplitudes; second, parameters selection based on satisfying the dynamic vibration absorber condition that can avoid the anti-resonance between the two peaks while maximizing the response in between those two peaks; third, proper selection of the distance between the two magnets to widen the bandwidth through adding the nonlinearity to the system. Simulations for the proposed harvester are carried out using MATLAB/Simulink over a wide bandwidth of frequencies. The results show that the proposed harvester can generate adequate voltage of 5 volts across a load of 1 GΩ over a wide range of frequency from 26 to 35 Hz for the proper distance between the two magnets that is based on the system parameters. Also, another frequency range from 5 to 11 Hz is targeted to insure that the proposed design methodology can be generalized.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"324 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124581234","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 : 2017-11-01DOI: 10.1109/MFI.2017.8170423
Carlos A. Garcia, Esteban X. Castellanos, Jorge Buele, J. Espinoza, Carmen Beltran, M. Pilatásig, Eddie D. Galarza, Marcelo V. García
In the present day, industry has been brought closer to the Industry 4.0 paradigm due to its needs of reaching better industrial communications and improving the control process. According to the IIoT concepts, Cyber Physical Production Systems (CPPS) are normally connected to each other, and also connected to the virtual world of global digital networks. By introducing the IEC-61499 standard, CPPS will be allowed to implement flexible, reconfigurable, and distributed controllers. Fuzzy controllers are an advanced control solution that adds certain level of human reasoning to the system, making it possible to obtain a suitable alternative for traditional controllers. Due to the requirement of IEC-61499 of using as controller device an embedded platform, the encapsulation of advanced techniques of control in the algorithms of a systems becomes more simple and efficient. This is why it is necessary to provide the industry with low-cost alternatives who can easily integrate more complex and efficient controllers, making it possible to redirect the CPPS development to a new variety of devices. This paper proposes the development of the Function Blocks (FBs) needed in order to create a distributed control system based on Fuzzy Logic to control an analog process.
{"title":"Fuzzy control implementation in low cost CPPS devices","authors":"Carlos A. Garcia, Esteban X. Castellanos, Jorge Buele, J. Espinoza, Carmen Beltran, M. Pilatásig, Eddie D. Galarza, Marcelo V. García","doi":"10.1109/MFI.2017.8170423","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170423","url":null,"abstract":"In the present day, industry has been brought closer to the Industry 4.0 paradigm due to its needs of reaching better industrial communications and improving the control process. According to the IIoT concepts, Cyber Physical Production Systems (CPPS) are normally connected to each other, and also connected to the virtual world of global digital networks. By introducing the IEC-61499 standard, CPPS will be allowed to implement flexible, reconfigurable, and distributed controllers. Fuzzy controllers are an advanced control solution that adds certain level of human reasoning to the system, making it possible to obtain a suitable alternative for traditional controllers. Due to the requirement of IEC-61499 of using as controller device an embedded platform, the encapsulation of advanced techniques of control in the algorithms of a systems becomes more simple and efficient. This is why it is necessary to provide the industry with low-cost alternatives who can easily integrate more complex and efficient controllers, making it possible to redirect the CPPS development to a new variety of devices. This paper proposes the development of the Function Blocks (FBs) needed in order to create a distributed control system based on Fuzzy Logic to control an analog process.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124402356","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 : 2017-11-01DOI: 10.1109/MFI.2017.8170442
Song-Mi Lee, Heeryon Cho, S. Yoon
Noise and variability in accelerometer data collected using smart devices obscure accurate human activity recognition. In order to tackle the degradation of the triaxial accelerometer data caused by noise and individual user differences, we propose a statistical noise reduction method using total variation minimization to attenuate the noise mixed in the magnitude feature vector generated from triaxial accelerometer data. The experimental results using Random Forest classifier prove that our noise removal approach is constructive in significantly improving the human activity recognition performance.
{"title":"Statistical noise reduction for robust human activity recognition","authors":"Song-Mi Lee, Heeryon Cho, S. Yoon","doi":"10.1109/MFI.2017.8170442","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170442","url":null,"abstract":"Noise and variability in accelerometer data collected using smart devices obscure accurate human activity recognition. In order to tackle the degradation of the triaxial accelerometer data caused by noise and individual user differences, we propose a statistical noise reduction method using total variation minimization to attenuate the noise mixed in the magnitude feature vector generated from triaxial accelerometer data. The experimental results using Random Forest classifier prove that our noise removal approach is constructive in significantly improving the human activity recognition performance.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116053297","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 : 2017-11-01DOI: 10.1109/MFI.2017.8170363
Won-Seok Choi, Dongweon Yoon, TaeEon Kim, Jangmyung Lee
In this paper we propose oil skimmer monitoring system by user-controlled smart phone IoT(Internet of Things)-specific controller and built in BLE 4.0. The device Hardware of the IoT is the trend of OSHW(Open Source Hardware). This opens up free design content for physical artifacts together spirit with FOSS (Free and Open Source Software). As a result of the derivation of Open-source culture making an ecosystem, it is shared and developed by the majority. Software(Firmware, OS, Application) develop a standard reference board and by providing relevant sources of a trend in OSHW that emphasizes accessibility or ease of implementation for development.
{"title":"Oil skimmer and controller monitoring system using IoT technology","authors":"Won-Seok Choi, Dongweon Yoon, TaeEon Kim, Jangmyung Lee","doi":"10.1109/MFI.2017.8170363","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170363","url":null,"abstract":"In this paper we propose oil skimmer monitoring system by user-controlled smart phone IoT(Internet of Things)-specific controller and built in BLE 4.0. The device Hardware of the IoT is the trend of OSHW(Open Source Hardware). This opens up free design content for physical artifacts together spirit with FOSS (Free and Open Source Software). As a result of the derivation of Open-source culture making an ecosystem, it is shared and developed by the majority. Software(Firmware, OS, Application) develop a standard reference board and by providing relevant sources of a trend in OSHW that emphasizes accessibility or ease of implementation for development.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131864484","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 : 2017-11-01DOI: 10.1109/MFI.2017.8170387
Do Young Kim, D. Yun
An interesting about MAV and a research for flapping wing aero vehicle has been increased. It is required that take-off and landing abilities to lift up a vehicle by flapping motion to achieve a special purpose for flapping wing aero vehicle. An airfoil has to generate lift force more than a weight of the body to work take-off from the ground without additional force. MAV can generate enough lift force, abilities of MAV: load capacity, reliability, damage reduction. In this study, we propose a four-bar linkage structure to generate lift force. And an experiment will be set up to measure the lift force with a folding mechanism of flapping wing. After this, the result will be used for development of nature-inspired bird robot.
{"title":"The effect of folding on motion flapping wing aero vehicle","authors":"Do Young Kim, D. Yun","doi":"10.1109/MFI.2017.8170387","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170387","url":null,"abstract":"An interesting about MAV and a research for flapping wing aero vehicle has been increased. It is required that take-off and landing abilities to lift up a vehicle by flapping motion to achieve a special purpose for flapping wing aero vehicle. An airfoil has to generate lift force more than a weight of the body to work take-off from the ground without additional force. MAV can generate enough lift force, abilities of MAV: load capacity, reliability, damage reduction. In this study, we propose a four-bar linkage structure to generate lift force. And an experiment will be set up to measure the lift force with a folding mechanism of flapping wing. After this, the result will be used for development of nature-inspired bird robot.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134151675","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 : 2017-11-01DOI: 10.1109/MFI.2017.8170453
Kolja Thormann, J. Honer, M. Baum
This paper presents a novel method to extract and track road boundaries in a temporal sequence of occupancy grids collected from a moving vehicle that is equipped with a laser scanner. The road boundaries are represented as circular arcs, where it is assumed that the boundaries are parallel to the driving direction. In order to find the optimal parameters of the circular arcs, first a one-dimensional optimization problem over the curvature is solved. Second, based on the optimal curvature, the optimal offset, i.e., radius, is determined. In order to obtain robust and smooth road boundary estimates, we suggest to employ a tracking algorithm, i.e., the Integrated Probabilistic Data Association (IPDA). The overall method is evaluated with real-world data from a highway scenario and compared with two state-of-the-art methods.
{"title":"Fast road boundary detection and tracking in occupancy grids from laser scans","authors":"Kolja Thormann, J. Honer, M. Baum","doi":"10.1109/MFI.2017.8170453","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170453","url":null,"abstract":"This paper presents a novel method to extract and track road boundaries in a temporal sequence of occupancy grids collected from a moving vehicle that is equipped with a laser scanner. The road boundaries are represented as circular arcs, where it is assumed that the boundaries are parallel to the driving direction. In order to find the optimal parameters of the circular arcs, first a one-dimensional optimization problem over the curvature is solved. Second, based on the optimal curvature, the optimal offset, i.e., radius, is determined. In order to obtain robust and smooth road boundary estimates, we suggest to employ a tracking algorithm, i.e., the Integrated Probabilistic Data Association (IPDA). The overall method is evaluated with real-world data from a highway scenario and compared with two state-of-the-art methods.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125727864","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 : 2017-11-01DOI: 10.1109/MFI.2017.8170434
Seong-Geun Shin, Dae-Ryong Ahn, Hyuck-Kee Lee
In object tracking field, occlusion situations between objects are important factors that degrade the performance of tracking algorithms. In this paper, we present a track management method in the tracking level to solve the discontinuous tracking problem caused by occlusions between detected objects. This work is performed by predicting the occlusion situation between detected objects and managing the state of tracks based on an approach to track-to-track fusion in a high-level sensor fusion approach using a lidar and a monocular camera sensor. The occlusion prediction is computed by taking into account the width, length, position and azimuth angle of the detected objects. The track management system manages the occlusion state of the track from the result of occlusion prediction as well as the initialization, creation, confirmation and deletion of the tracks. The proposed approach has been verified in the occlusion situation between pedestrians, and our experimental results showed the intended performance in the occlusion situation between pedestrians.
{"title":"Occlusion handling and track management method of high-level sensor fusion for robust pedestrian tracking","authors":"Seong-Geun Shin, Dae-Ryong Ahn, Hyuck-Kee Lee","doi":"10.1109/MFI.2017.8170434","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170434","url":null,"abstract":"In object tracking field, occlusion situations between objects are important factors that degrade the performance of tracking algorithms. In this paper, we present a track management method in the tracking level to solve the discontinuous tracking problem caused by occlusions between detected objects. This work is performed by predicting the occlusion situation between detected objects and managing the state of tracks based on an approach to track-to-track fusion in a high-level sensor fusion approach using a lidar and a monocular camera sensor. The occlusion prediction is computed by taking into account the width, length, position and azimuth angle of the detected objects. The track management system manages the occlusion state of the track from the result of occlusion prediction as well as the initialization, creation, confirmation and deletion of the tracks. The proposed approach has been verified in the occlusion situation between pedestrians, and our experimental results showed the intended performance in the occlusion situation between pedestrians.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"95 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131660432","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 : 2017-11-01DOI: 10.1109/MFI.2017.8170437
Katharina Dormann, B. Noack, U. Hanebeck
The centralized Kalman filter can be implemented in such a way that the required calculations can be distributed over multiple nodes in a network, each of which processes only the locally acquired sensor data. The main downside of this implementation is that it requires each distributed sensor node to communicate with the fusion center in every time step so as to compute the optimal state estimate. In this paper, two distributed Kalman filtering algorithms are proposed to overcome these limitations. The first algorithm merely requires communication of each local sensor node with the fusion center in every other time step. The second algorithm even allows for a lower communicate rate. Both algorithms apply event-based communication to compute consistent estimates and to reduce the estimation error for a fixed communication rate. Simulations demonstrate that both algorithms perform better in terms of the mean squared estimation error than the centralized Kalman filter.
{"title":"Distributed Kalman filtering with reduced transmission rate","authors":"Katharina Dormann, B. Noack, U. Hanebeck","doi":"10.1109/MFI.2017.8170437","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170437","url":null,"abstract":"The centralized Kalman filter can be implemented in such a way that the required calculations can be distributed over multiple nodes in a network, each of which processes only the locally acquired sensor data. The main downside of this implementation is that it requires each distributed sensor node to communicate with the fusion center in every time step so as to compute the optimal state estimate. In this paper, two distributed Kalman filtering algorithms are proposed to overcome these limitations. The first algorithm merely requires communication of each local sensor node with the fusion center in every other time step. The second algorithm even allows for a lower communicate rate. Both algorithms apply event-based communication to compute consistent estimates and to reduce the estimation error for a fixed communication rate. Simulations demonstrate that both algorithms perform better in terms of the mean squared estimation error than the centralized Kalman filter.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"87 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115816100","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 : 2017-11-01DOI: 10.1109/MFI.2017.8170367
M. Atif, Sukhan Lee
Structured light 3D camera systems are composed of a commercial video projector and a machine vision camera. High scan speed structured light 3D camera systems, suffers synchronization problems due to a mismatch in the projector screen refresh rate, and camera capturing speed. Commercial video projectors project with fixed refresh rate, frame rate of machine vision camera's increase or decrease with the resolution. Synchronization between projected frame and captured frame cannot be achieved through computer graphic's interface, due to limited control of hardware of computer. This paper presents a method to project structured light patterns and trigger for the camera according to camera frame rate and projector screen refresh rate. An adaptive frame rate pattern projection framework is implemented on Field Programmable Gate Array (FPGA), to achieve camera projector synchronization at any camera frame rate and projector refresh rate, which improves the accuracy of the point cloud.
{"title":"Adaptive frame rate pattern projection for structured light 3D camera system","authors":"M. Atif, Sukhan Lee","doi":"10.1109/MFI.2017.8170367","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170367","url":null,"abstract":"Structured light 3D camera systems are composed of a commercial video projector and a machine vision camera. High scan speed structured light 3D camera systems, suffers synchronization problems due to a mismatch in the projector screen refresh rate, and camera capturing speed. Commercial video projectors project with fixed refresh rate, frame rate of machine vision camera's increase or decrease with the resolution. Synchronization between projected frame and captured frame cannot be achieved through computer graphic's interface, due to limited control of hardware of computer. This paper presents a method to project structured light patterns and trigger for the camera according to camera frame rate and projector screen refresh rate. An adaptive frame rate pattern projection framework is implemented on Field Programmable Gate Array (FPGA), to achieve camera projector synchronization at any camera frame rate and projector refresh rate, which improves the accuracy of the point cloud.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117148222","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 : 2017-11-01DOI: 10.1109/MFI.2017.8170441
Inhwan Hwang, Geonho Cha, Songhwai Oh
Human action recognition has been studied in many fields including computer vision and sensor networks using inertial sensors. However, there are limitations such as spatial constraints, occlusions in images, sensor unreliability, and the inconvenience of users. In order to solve these problems we suggest a sensor fusion method for human action recognition exploiting RGB images from a single fixed camera and a single wrist mounted inertial sensor. These two different domain information can complement each other to fill the deficiencies that exist in both image based and inertial sensor based human action recognition methods. We propose two convolutional neural network (CNN) based feature extraction networks for image and inertial sensor data and a recurrent neural network (RNN) based classification network with long short term memory (LSTM) units. Training of deep neural networks and testing are done with synchronized images and sensor data collected from five individuals. The proposed method results in better performance compared to single sensor-based methods with an accuracy of 86.9% in cross-validation. We also verify that the proposed algorithm robustly classifies the target action when there are failures in detecting body joints from images.
{"title":"Multi-modal human action recognition using deep neural networks fusing image and inertial sensor data","authors":"Inhwan Hwang, Geonho Cha, Songhwai Oh","doi":"10.1109/MFI.2017.8170441","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170441","url":null,"abstract":"Human action recognition has been studied in many fields including computer vision and sensor networks using inertial sensors. However, there are limitations such as spatial constraints, occlusions in images, sensor unreliability, and the inconvenience of users. In order to solve these problems we suggest a sensor fusion method for human action recognition exploiting RGB images from a single fixed camera and a single wrist mounted inertial sensor. These two different domain information can complement each other to fill the deficiencies that exist in both image based and inertial sensor based human action recognition methods. We propose two convolutional neural network (CNN) based feature extraction networks for image and inertial sensor data and a recurrent neural network (RNN) based classification network with long short term memory (LSTM) units. Training of deep neural networks and testing are done with synchronized images and sensor data collected from five individuals. The proposed method results in better performance compared to single sensor-based methods with an accuracy of 86.9% in cross-validation. We also verify that the proposed algorithm robustly classifies the target action when there are failures in detecting body joints from images.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127073773","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：