Guided waves are an attractive tool for monitoring large structures. They are able to interrogate sizable areas at once, are sensitive to structural damage, and carry information about the damage location. Yet, there are many challenges to guided wave data analysis. First, guided waves are represented by a complex set of dispersive wave modes that distort their shape and phase as they travel through the medium. Second, the complex properties of guided waves are often unknown. Third, the unknown properties often vary as a function of time. As a result, to effectively analyze guided waves, their complex properties must be learned, tracked, and leveraged. This paper describes signal processing tools for managing the complexity, uncertainty, and variability of guided waves for structural health monitoring applications. We use a stretch-based and a matrix decomposition based temperature compensation methods to handle variability, sparse wavenumber analysis to learn the uncertain properties of guided waves, and matched field processing to then leverage these complexities. We demonstrate how these methods are combined for effective guided wave data analysis with experimental data from an aluminum plate under varying temperature conditions.
The pressure change method of leak testing has the serious disadvantage of being susceptible to ambient temperature change. Two or more successive leak tests are typically performed to eliminate the error caused by such change. We propose a new leak detection scheme in which multiple successive leak tests are performed on the basis of the Lagrange interpolation formula. With this scheme, we can optimally perform multiple successive leak tests for a given temperature variation. After describing the theoretical aspects of the proposed scheme, we will show experimental results of leak detection using a prototype leak detector and model piping. Results indicate that a leak can be detected using the proposed scheme without being affected by the given temperature variation.
In this paper, we propose a method for detecting moving objects using a moving stereo camera. First, the camera motion parameters are estimated by using optical flow with a stereo camera. Second, the optical flow occurring in the background is removed. Finally, moving objects are detected individually by labeling the remaining optical flow. The proposed method is evaluated through experiments using two pedestrians in an indoor environment.
Unmanned aerial vehicle (UAV)-assisted microwave wireless power transfer (WPT) enables the deployment of radio frequency (RF)-power-driven, battery-less sensor nodes in rural areas such as farm fields. Unlike in urban areas, suitable ambient radio waves for RF energy harvesting are not available in rural areas, therefore, we propose using UAVs to carry active RF sources. The UAVs will cover large fields and function as movable power feeders. To feed power to sensor nodes or any power feeding target efficiently, a UAV needs to be navigated to the desired power feeding point, which is usually right above the target; however, a Global Positioning Services (GPS) system is not accurate enough for this purpose. Therefore, this paper presents two trilateration-inspired sensor node position estimation methods for UAV WPT, based on the relationship between the sensor-UAV distance and power transmission efficiency. After the UAV collects data of several distances, measured by the sensor node from different UAV locations, the UAV then estimates the position of a sensor node by utilizing the data. In the direction-based approach, circles with the radius set as the measured distance are first centered at the measurement positions. Then, all the intersections of the two circles are calculated. Further, by relying on the assumption that the sensor node would be in the direction where the largest number of intersections is observed, the average position of the intersections included in the direction is regarded as the estimated position of the sensor. In the least squares approach, the position that minimizes the sum of squares of errors, obtained from the measurement results, is assumed to be the sensor position. By comparing the direction-based and least squares approaches to the conventional hill climbing method, we found that the least squares and direction-based approaches can complete power feeding faster in average by 52% and 26%, respectively, compared to the hill climbing method. Combining GPS with our least squares approach will enable the UAV to reach the appropriate zone rapidly and complete the power supply process quickly so that power may be delivered to more sensor nodes in less time.
World health statistics about overweight and obesity show that overweight, obesity and diet-related diseases still remain major health risks. According to the World Health Organization, most of the world's population live in countries where overweight and obesity kill more people than underweight. Recently, many studies have been driven by the motivation of elaborating the “ideal” solutions to prevent and/or monitor overweight, obesity and diet-related diseases, and to encourage healthy diet and lifestyle. In this work, we present our idea to automatically measure food weight and calories, from food photo using ordinary chopsticks as a measurement reference. The analysis of the obtained results show that the use of near-by eating utensils combined with computer vision techniques is a great and exploitable approach to ubiquitously help in diet assessment and obesity treatment.
This paper proposes a monocular vision based obstacle detection algorithm for autonomous mobile robots. Our main algorithm consists of two stages. In the first stage, we use an inverse perspective mapping (IPM) based method for detecting small portions of an obstacle in the input image. In the second stage, we perform image abstraction and geodesic distance computation for segmenting the obstacle. We use the simple linear iterative clustering (SLIC) superpixel algorithm for decomposing the image into basic elements that preserve relevant structure, but abstract undesirable detail. The source superpixel for geodesic distance computation is selected according to semi-local texture features. Then we compute the obstacle score for accurate segmentation. Experimental results have shown that our proposed method achieves accuracy comparable to the state-of-the-art method, with more than 7 times faster computation.
A Doppler sensor enables us to detect the movement of objects. It transmits an electro-magnetic wave and receives signals yielded by the reflection at the objects so that we can detect the movement of the objects by measuring the frequency deviation between the transmitted- and the received signals. Recently, it is also drawing attention as a tool of non-contact vital sensing such as heartbeats and breathing. This is particularly interesting as a means for monitoring of patients in hospitals or residents in nursing homes in order to make such facilities more efficient. In this paper, aiming at such application, we propose a new very simple signal processing scheme in order to estimate (1) the number of signals, (2) the period of each signal, and (3) the fundamental waveform. It is also emphasized that the estimations (1)-(3) are available for two signals. The performance is verified through computer simulations.
These days, smart home applications such as concierge service for residents, home appliance control, and so on are attracting attention. To realize these applications, we need a system which recognizes various human activities accurately with a low cost device. There are many studies on the activity recognition in a smart home. We also have proposed an activity recognition technique in a smart home by utilizing digital-output-PIR (passive infrared) sensors, door sensors, and power meters. However, the study has an unsolved issue: we cannot distinguish similar activities happening at the same place, for example, “eating” and “reading” while sitting on a sofa. In order to cope with this challenge, we introduce ALPAS: analog-output-PIR-sensor-based activity recognition technique which recognizes the different activities in the same place. Our technique recognizes user's activity by utilizing machine learning with frequency components of the sensor's output as features. However, because the number of features used in ALPAS is 1000 for each analog PIR-sensor, a large capacity memory is required. To reduce the number of features, we select a part of the sensing data. We call the starting point of the selected data as starting frequency (SF) and ending part as ending frequency (EF). We searched SF and EF using a grid search, and evaluated the recognition accuracy. We evaluated the proposed technique in a smarthome testbed. In the evaluation, five participants performed four different activities while sitting on a sofa. As a result, we achieved F-Measure: 63.9% when the EF is 1.4Hz, and F-Measure: 50% or lower when the SF is 9.9Hz or higher.
Road maintenance requires local city governments to dedicate a substantial amount of funds in finding and repairing damaged traffic marks and pavements. In developed cities, the total road length is so large that the cost becomes unreasonably high. In this paper, we propose a method of sensing damaged traffic marks from images captured by a camera mounted to a car, for the purpose of reducing road maintenance cost. In particular, we utilized convolutional neural networks (CNN), as well as linear support vector machines (SVM) and Random Forest, in developing a system of damage detection. The experiments used thousands of images captured in the wild and showed that the method can detect damages using CNN with 93% accuracy, at maximum, and at reasonable speed (55 images per second).
To successfully reconstruct a three-dimensional (3D) model from images, it is necessary to make the user aware of whether sufficient data has been obtained. We propose a novel approach to detect defective regions in 3D reconstruction during image acquisition. Our method uses line-based segmentation to segment the acquired images into structural and non-structural regions. Then, using the 3D structures derived from these images, defects detected in the structural region are used to suppress spurious artifacts. Visualization of a defective region in a newly acquired image allows the user to comprehend the reconstruction state and adjust image acquisition. The proposed method was experimentally demonstrated to work successfully in a range of outdoor environments.
This paper addresses the problem of energy-efficient power assist control for quasiperiodic motions. The simplest assist method would be to apply additional torque in proportion to the instantaneous value of torque generated by a user. In our previous study, it was shown that energy efficiency improves by flattening the torque pattern. To cope with the frequency fluctuation of our motion, we introduce a periodic disturbance observer with a frequency estimator and suppress the pulsation of the human torque based on a disturbance observer framework. The effectiveness of the proposed method is evaluated through numerical simulations and experiments with an actual electric bicycle being pedaled by a human.
Aging is a serious issue in our global society. In particular, a smaller working population increases the burden on nurses and other healthcare professionals. To reduce the monitoring burden on nursing, a system that enables quantitative monitoring of the elderly automatically in their home is very desirable. We have proposed a novel system to classify the presence or absence of a subject within a designated area using a single microwave Doppler sensor. The proposed system utilizes the respirational signal obtained from the sensor and then classified using an SVM. In this paper, we discuss feature components that would positively or negatively affect the performance of the system. The experimental results show that the combination of a demodulated time domain feature and frequency domain features affect positively to the accuracy of the system.
Rapid adapting type-I (RA-I) receptor is one type of mechanoreceptors in the human skin. They are believed to be responsible for the detection of stimuli that produce minute skin motion (flutter, slip, microgeometric surface features). The neurophysiological experiments in the paper [J.R. Phillips et al. J. Neurophysiol., Vol. 46, pp. 1192-1203, 1981]raise a question about why the RA-I afferent (innervated into RA-I receptor) fails to represents the stimulus with the width less than 3mm and why their response is anisotropy. It is unclear whether the skin's mechanics or the specific afferent branching of mechanoreceptors themselves are accounted for these phenomena. The present work seeks an interpretation of the neurophysiological phenomena, using a biomechanical finite-element (FE) model with a transduction sub-layer and synthetic sub-model for afferent current. The predicted afferent current matched well with the neural recordings in previous reports. This result suggests a major role of afferent branching in regard to the neurophysiological phenomena.
This paper proposes a novel sit-to-stand and stand-to-sit (STS) training system for hemiplegic patients with clonus. Most hemiplegic patients depend on the unaffected leg and perform plantar flexion involuntarily during STS movements. The developed system integrates a visual biofeedback method with a movement assistance mechanism. The system provides users' ground reaction force (GRF) through a monitor and supports STS movements by moving the seat. We also proposed STS training with physical inhibition by using the developed orthosis to manage clonus. We performed an experiment on a hemiplegic patient who experienced clonus in the affected leg to verify that the user was able to increase the GRF of the heel part (GRFh) of the affected leg while performing STS movements using the system and that the system employing the orthosis reduced clonus. The results showed that GRFh in the intervention test that employed the proposed system was greater than that in the baseline test without the system. We also verified that a significant difference existed between the baseline and intervention test results (p<0.05). In addition, when the orthosis was used, the participant was able to prevent muscle cramps resulting from the clonus by avoiding the contraction of the affected leg muscles. The results indicated that STS training using the system helped the user to increase GRFh and reduce the clonus. In addition, from the viewpoint of motor learning, we deduced that an integrated assistive system with informatics-based biofeedback and physical human-machine interaction would be effective in reconstructing the feedback loop in users' central nervous system.
This paper describes the structural analysis of the programming operation of the tangible programming tool P-CUBE, which is a tangible block-type programming tool that was developed for use by beginners, including those with visual impairments. An experiment was conducted to compare P-CUBE and conventional programming software platforms, and it was demonstrated that P-CUBE encourages the development of beginners' understanding of the structure of a program. The differences between the programming operations of P-CUBE and the programming software are considered to be a factor causing differences in users' understanding of programming structures. The structures of the programming operations were analyzed using formal concept analysis, and differences between the logical frameworks in the operation of P-CUBE and a conventional software platform were demonstrated.
Model Driven Architecture (MDA) is a program development method designed for experts in modeling, but not in programming, and it has not been widely accepted yet. In this paper, we propose a new program development method that enables modeling experts to develop a program based on the dynamic model description in accordance with the modeling method in their domain. Furthermore, we discuss the structure of a programming language specialized for this modeling method and the modeling environment to develop this new program development method.
Automation systems that regard humans as the final authority have been found to be efficient and are therefore widely accepted. However, it has been argued that automation needs to be allowed to act autonomously in some time-critical situations, such as road traffic accidents. Possible interactions might occur between humans, especially those not well trained as car drivers, and authorised autonomous systems. A study using a driving simulator was designed to examine human-machine interactions when driving with two types of assistance systems: sharing of steering control that provides haptic control guidance through the steering wheel to resist hazardous lane changes, and an automatic cooperative system that acts autonomously to avoid hazardous lane changes. Whilst the drivers were in charge of steering in all circumstances when sharing the steering control, they were unable to steer their vehicles during the autonomous control. Results showed that increasing automation authority does not necessarily lead to improved safety. Other factors like human-machine cooperation need to be considered when the assistance system experiences functional limitations. Although lane change crashes were significantly reduced when the drivers were supported by the autonomous system, drivers reacted earlier and more convenient when supported by the haptic system.
Design methods for control systems based on plant models have been developed for many years. If a mathematical model is accurately obtained from the input-output relation of a plant, then the designed controller for the model performs well for the control system connected with the plant. However, the desired control performance might not be achieved when there is an undeniable modeling error. To overcome this problem, the authors proposed the model error compensator (MEC) to minimize the effect of the modeling error between the plant and the model. The MEC works well for many control systems, such as unstable systems and non-linear systems. However applying the MEC to non-minimum-phase plants is difficult because of their control system structures. Non-minimum-phase plants are well known for being difficult to control. This paper proposes an MEC with a parallel feed-forward filter (PFF). The PFF is used to cancel the non-minimum-phase characteristics of the plant. The effectiveness of the proposed method is illustrated through numerical examples.
Stereo vision is a well-known technique for vision-based 3D reconstruction of environments. Recently developed spherical cameras can be used to extend the concept to all 360° and provide LIDAR-like 360 degree 3D data with color information. In order to perform accurate stereo disparity estimation, the accurate relative pose between the two cameras, represented by the five degree of freedom epipolar geometry, needs to be known. However, it is always tedious to mechanically align and/or calibrate such systems. We propose a technique to recover the complete five degree of freedom parameters of the epipolar geometry in a single minimization with a dense approach involving all the individual pixel displacements (optical flow) between two camera views. Taking advantage of the spherical image geometry, a non-linear least squares optimization based on the dense optical flow directly minimizes the angles between pixel displacements and epipolar curves in order to align them. This approach is particularly suitable for dense 3D reconstruction as the pixel-to-pixel disparity between the two images can be calculated accurately and converted to a dense point cloud. Further, there are no assumptions about the direction of camera displacement. We demonstrate this method by showing some error evaluations, examples of successfully rectified spherical stereo pairs, and the dense 3D models generated from them.