Pub Date : 2017-11-01DOI: 10.1109/MFI.2017.8170399
K. Kwon, Seung Hyun Lee, M. Y. Kim
Image-to-patient registration process is required to use actively pre-operative images such as CT and MRI during operation for surgical navigation system. One method to utilize scanning data of patients and 3D data from MRI or CT images is dealt with in this paper. After 3D surface measurement device measures the surface of patient's surgical site, this 3D data is registered to CT or MRI data using computer-based optimization algorithms like conventional ICP algorithms. However, general ICP algorithm has some disadvantages that it takes a long converging time if a proper initial location is not set up and also suffers from local minimum problem during the process. In this paper, we propose an automatic image-to-patient registration method that can accurately find a proper initial location without manual intervention of surgical operators. The proposed method finds and extracts the initial starting location for ICP by converting 3D data set of MRI or CT images and surface scanning data to 2D curvature images and by performing H-K curvature image matching between them automatically. It is based on the characteristics that curvature features are robust to the rotation, translation and even some deformation. Automatic image-to-patient registration is implemented by precisely 3D registration the extracted CT ROI and the patient's surface measurement data using ICP algorithm.
{"title":"A patient-to-CT registration method based on spherical unwrapping and H-K curvature descriptors for surgical navigation system","authors":"K. Kwon, Seung Hyun Lee, M. Y. Kim","doi":"10.1109/MFI.2017.8170399","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170399","url":null,"abstract":"Image-to-patient registration process is required to use actively pre-operative images such as CT and MRI during operation for surgical navigation system. One method to utilize scanning data of patients and 3D data from MRI or CT images is dealt with in this paper. After 3D surface measurement device measures the surface of patient's surgical site, this 3D data is registered to CT or MRI data using computer-based optimization algorithms like conventional ICP algorithms. However, general ICP algorithm has some disadvantages that it takes a long converging time if a proper initial location is not set up and also suffers from local minimum problem during the process. In this paper, we propose an automatic image-to-patient registration method that can accurately find a proper initial location without manual intervention of surgical operators. The proposed method finds and extracts the initial starting location for ICP by converting 3D data set of MRI or CT images and surface scanning data to 2D curvature images and by performing H-K curvature image matching between them automatically. It is based on the characteristics that curvature features are robust to the rotation, translation and even some deformation. Automatic image-to-patient registration is implemented by precisely 3D registration the extracted CT ROI and the patient's surface measurement data using ICP algorithm.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"75 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":"127263179","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}
Pub Date : 2017-11-01DOI: 10.1109/MFI.2017.8170420
Md. Zia Uddin, W. Khaksar, J. Tørresen
In this work, we propose a novel human activity recognition method from depth videos using robust spatiotemporal features with convolutional neural network. From the depth images of activities, human body parts are segmented based on random features on a random forest. From the segmented body parts in a depth image of an activity video, spatial features are extracted such as angles of the 3-D body joint pairs, means and variances of the depth information in each part of the body. The spatial features are then augmented with the motion features such as magnitude and direction of joints in next image of the video. Finally, the spatiotemporal features are applied to a convolutional neural network for activity training and recognition. The deep learning-based activity recognition method outperforms other traditional methods.
{"title":"Human activity recognition using robust spatiotemporal features and convolutional neural network","authors":"Md. Zia Uddin, W. Khaksar, J. Tørresen","doi":"10.1109/MFI.2017.8170420","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170420","url":null,"abstract":"In this work, we propose a novel human activity recognition method from depth videos using robust spatiotemporal features with convolutional neural network. From the depth images of activities, human body parts are segmented based on random features on a random forest. From the segmented body parts in a depth image of an activity video, spatial features are extracted such as angles of the 3-D body joint pairs, means and variances of the depth information in each part of the body. The spatial features are then augmented with the motion features such as magnitude and direction of joints in next image of the video. Finally, the spatiotemporal features are applied to a convolutional neural network for activity training and recognition. The deep learning-based activity recognition method outperforms other traditional methods.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"5 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":"122759030","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.2017.8170393
Xudong Sun, F. Sun, Bin Wang, Jianqin Yin, Xiaolin Sheng, Qinghua Xiao
Frontier-based exploration is the most common approach to exploration, a fundamental problem in robotics. Laser scanner and Kinect have been widely used in robotic application for simultaneous localization and mapping (SLAM) separately. The paper proposes a method to combine the data from Kinect and laser scanner to perform a Frontier-based exploration SLAM. The 2 sensors will be installed facing forward and facing backward in opposite directions which make robot have wider vision, thus the robot can detect more complex surrounding features to increase the exploration efficiency and to construct a more accurate map of the unknown environment.
{"title":"Robotic autonomous exploration SLAM using combination of Kinect and laser scanner","authors":"Xudong Sun, F. Sun, Bin Wang, Jianqin Yin, Xiaolin Sheng, Qinghua Xiao","doi":"10.1109/MFI.2017.8170393","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170393","url":null,"abstract":"Frontier-based exploration is the most common approach to exploration, a fundamental problem in robotics. Laser scanner and Kinect have been widely used in robotic application for simultaneous localization and mapping (SLAM) separately. The paper proposes a method to combine the data from Kinect and laser scanner to perform a Frontier-based exploration SLAM. The 2 sensors will be installed facing forward and facing backward in opposite directions which make robot have wider vision, thus the robot can detect more complex surrounding features to increase the exploration efficiency and to construct a more accurate map of the unknown environment.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"8 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":"130886901","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 paper, we introduce a new hardware platform that mimics a compound eye of an insect and propose an algorithm to detect objects using it. The compound eye camera has a wide viewing angle and simulates a number of single eyes on its hemisphere. Each single eye is an elementary unit to acquire visual inputs. Visual information from single eyes is hierarchically merged to estimate objectness. We achieve the accuracy of 77.14% on a combined dataset of PASCAL VOC 2012 and COCO-Stuff 10K databases.
{"title":"Estimating objectness using a compound eye camera","authors":"Hwiyeon Yoo, Donghoon Lee, Geonho Cha, Songhwai Oh","doi":"10.1109/MFI.2017.8170418","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170418","url":null,"abstract":"In this paper, we introduce a new hardware platform that mimics a compound eye of an insect and propose an algorithm to detect objects using it. The compound eye camera has a wide viewing angle and simulates a number of single eyes on its hemisphere. Each single eye is an elementary unit to acquire visual inputs. Visual information from single eyes is hierarchically merged to estimate objectness. We achieve the accuracy of 77.14% on a combined dataset of PASCAL VOC 2012 and COCO-Stuff 10K databases.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"46 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":"123854735","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.2017.8170373
Soumyo Das, Yamini Yarlagadda, Prashantkumar B. Vora, Sabarish R. P. Nair
This paper discusses about the trajectory planning and controlled maneuvering of the vehicle in parking assist mode. The objective of the proposed algorithm is to use low-cost hardware like ultrasonic sensor in order to provide automated parking assist to the vehicle. The concept of grid occupancy is formulated in free space detection algorithm to compute lateral-longitudinal grid cell vacancy. The vehicle will enable automated parking assist mode for controlled maneuvering on completion of free space detection followed by path planning. The reference planned path is an optimized trajectory with single maneuver for the vehicle to traverse in a free parking space. The localized trajectory and control algorithm of the perpendicular parking have been designed with reference to perception sensor input. The controller facilitates the intelligent navigation of vehicle based on measured obstacle free clearance distance and reference point. The steering control algorithm based on fuzzy logic is designed to provide an optimized maneuvering of the host vehicle in perpendicular parking space. In order to ease parking effort, an innovative approach of combined feedback and feed-forward based fuzzy controller has been illustrated. The controller performance has been evaluated in simulation environment keeping vehicle dynamics model in loop. The test scenario has been modeled in Carmaker to substantiate the optimization of route selection and a smooth transition of vehicle in parking assist mode during maneuvering into an empty parking space.
{"title":"Trajectory planning and fuzzy control for perpendicular parking","authors":"Soumyo Das, Yamini Yarlagadda, Prashantkumar B. Vora, Sabarish R. P. Nair","doi":"10.1109/MFI.2017.8170373","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170373","url":null,"abstract":"This paper discusses about the trajectory planning and controlled maneuvering of the vehicle in parking assist mode. The objective of the proposed algorithm is to use low-cost hardware like ultrasonic sensor in order to provide automated parking assist to the vehicle. The concept of grid occupancy is formulated in free space detection algorithm to compute lateral-longitudinal grid cell vacancy. The vehicle will enable automated parking assist mode for controlled maneuvering on completion of free space detection followed by path planning. The reference planned path is an optimized trajectory with single maneuver for the vehicle to traverse in a free parking space. The localized trajectory and control algorithm of the perpendicular parking have been designed with reference to perception sensor input. The controller facilitates the intelligent navigation of vehicle based on measured obstacle free clearance distance and reference point. The steering control algorithm based on fuzzy logic is designed to provide an optimized maneuvering of the host vehicle in perpendicular parking space. In order to ease parking effort, an innovative approach of combined feedback and feed-forward based fuzzy controller has been illustrated. The controller performance has been evaluated in simulation environment keeping vehicle dynamics model in loop. The test scenario has been modeled in Carmaker to substantiate the optimization of route selection and a smooth transition of vehicle in parking assist mode during maneuvering into an empty parking space.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"8 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":"122379887","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.2017.8170451
G. H. Lim, E. Pedrosa, F. Amaral, N. Lau, Artur Pereira, J. L. Azevedo, B. Cunha
This paper presents an integrated neural regularization method in fully-connected neural networks that jointly combines the cutting edge of regularization techniques; Dropout [1] and DropConnect [2]. With a small number of data set, trained feed-forward networks tend to show poor prediction performance on test data which has never been introduced while training. In order to reduce the overfitting, regularization methods commonly use only a sparse subset of their inputs. While a fully-connected layer with Dropout takes account of a randomly selected subset of hidden neurons with some probability, a layer with DropConnect only keeps a randomly selected subset of connections between neurons. It has been reported that their performances are dependent on domains. Image classification results show that the integrated method provides more degrees of freedom to achieve robust image recognition in the test phase. The experimental analyses on CIFAR-10 and one-hand gesture dataset show that the method provides the opportunity to improve classification performance.
{"title":"Neural regularization jointly involving neurons and connections for robust image classification","authors":"G. H. Lim, E. Pedrosa, F. Amaral, N. Lau, Artur Pereira, J. L. Azevedo, B. Cunha","doi":"10.1109/MFI.2017.8170451","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170451","url":null,"abstract":"This paper presents an integrated neural regularization method in fully-connected neural networks that jointly combines the cutting edge of regularization techniques; Dropout [1] and DropConnect [2]. With a small number of data set, trained feed-forward networks tend to show poor prediction performance on test data which has never been introduced while training. In order to reduce the overfitting, regularization methods commonly use only a sparse subset of their inputs. While a fully-connected layer with Dropout takes account of a randomly selected subset of hidden neurons with some probability, a layer with DropConnect only keeps a randomly selected subset of connections between neurons. It has been reported that their performances are dependent on domains. Image classification results show that the integrated method provides more degrees of freedom to achieve robust image recognition in the test phase. The experimental analyses on CIFAR-10 and one-hand gesture dataset show that the method provides the opportunity to improve classification performance.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"42 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":"127718349","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.2017.8170353
W. Müller, A. Kuwertz, D. Mühlenberg, J. Sander
In recent years, the usage of unmanned aircraft systems (UAS) for security-related purposes has increased, ranging from military applications to different areas of civil protection. The deployment of UAS can support security forces in achieving an enhanced situational awareness. However, in order to provide useful input to a situational picture, sensor data provided by UAS has to be integrated with information about the area and objects of interest from other sources. The aim of this study is to design a high-level data fusion component combining probabilistic information processing with logical and probabilistic reasoning, to support human operators in their situational awareness and improving their capabilities for making efficient and effective decisions. To this end, a fusion component based on the ISR (Intelligence, Surveillance and Reconnaissance) Analytics Architecture (ISR-AA) [1] is presented, incorporating an object-oriented world model (OOWM) for information integration, an expressive knowledge model and a reasoning component for detection of critical events. Approaches for translating the information contained in the OOWM into either an ontology for logical reasoning or a Markov logic network for probabilistic reasoning are presented.
{"title":"Semantic information fusion to enhance situational awareness in surveillance scenarios","authors":"W. Müller, A. Kuwertz, D. Mühlenberg, J. Sander","doi":"10.1109/MFI.2017.8170353","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170353","url":null,"abstract":"In recent years, the usage of unmanned aircraft systems (UAS) for security-related purposes has increased, ranging from military applications to different areas of civil protection. The deployment of UAS can support security forces in achieving an enhanced situational awareness. However, in order to provide useful input to a situational picture, sensor data provided by UAS has to be integrated with information about the area and objects of interest from other sources. The aim of this study is to design a high-level data fusion component combining probabilistic information processing with logical and probabilistic reasoning, to support human operators in their situational awareness and improving their capabilities for making efficient and effective decisions. To this end, a fusion component based on the ISR (Intelligence, Surveillance and Reconnaissance) Analytics Architecture (ISR-AA) [1] is presented, incorporating an object-oriented world model (OOWM) for information integration, an expressive knowledge model and a reasoning component for detection of critical events. Approaches for translating the information contained in the OOWM into either an ontology for logical reasoning or a Markov logic network for probabilistic reasoning are presented.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"65 3 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":"127553597","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.2017.8170401
Hendrik Vincent Koops, Kashish Garg, Munsung Kim, Jonathan Li, A. Volk, F. Franchetti
Improving trust in the state of Cyber-Physical Systems becomes increasingly important as more Cyber-Physical Systems tasks become autonomous. Research into the sound of Cyber-Physical Systems has shown that audio side-channel information from a single microphone can be used to accurately model traditional primary state sensor measurements such as speed and gear position. Furthermore, data integration research has shown that information from multiple heterogeneous sources can be integrated to create improved and more reliable data. In this paper, we present a multi-microphone machine learning data fusion approach to accurately predict ascending/hovering/descending states of a multi-rotor UAV in flight. We show that data fusion of multiple audio classifiers predicts these states with accuracies over 94%. Furthermore, we significantly improve the state predictions of single microphones, and outperform several other integration methods. These results add to a growing body of work showing that microphone side-channel approaches can be used in Cyber-Physical Systems to accurately model and improve the assurance of primary sensors measurements.
{"title":"Multirotor UAV state prediction through multi-microphone side-channel fusion","authors":"Hendrik Vincent Koops, Kashish Garg, Munsung Kim, Jonathan Li, A. Volk, F. Franchetti","doi":"10.1109/MFI.2017.8170401","DOIUrl":"https://doi.org/10.1109/MFI.2017.8170401","url":null,"abstract":"Improving trust in the state of Cyber-Physical Systems becomes increasingly important as more Cyber-Physical Systems tasks become autonomous. Research into the sound of Cyber-Physical Systems has shown that audio side-channel information from a single microphone can be used to accurately model traditional primary state sensor measurements such as speed and gear position. Furthermore, data integration research has shown that information from multiple heterogeneous sources can be integrated to create improved and more reliable data. In this paper, we present a multi-microphone machine learning data fusion approach to accurately predict ascending/hovering/descending states of a multi-rotor UAV in flight. We show that data fusion of multiple audio classifiers predicts these states with accuracies over 94%. Furthermore, we significantly improve the state predictions of single microphones, and outperform several other integration methods. These results add to a growing body of work showing that microphone side-channel approaches can be used in Cyber-Physical Systems to accurately model and improve the assurance of primary sensors measurements.","PeriodicalId":402371,"journal":{"name":"2017 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)","volume":"13 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":"123640958","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：