Bed-Exit Behavior Recognition for Real-Time Images within Limited Range

In the context of behavior recognition, the emerging bed-exit monitoring system demands a rapid deployment in the ward to support mobility and personalization. Mobility means the system can be installed and removed as required without construction; personalization indicates human body tracking is limited to the bed region so that only the target is monitored. To satisfy the above-mentioned requirements, the behavior recognition system aims to: (1) operate in a small-size device, typically an embedded system; (2) process a series of images with narrow fields of view (NFV) to detect bed-related behaviors. In general, wide-range images are preferred to obtain a good recognition performance for diverse behaviors, while NFV images are used with abrupt activities and therefore fit single-purpose applications. This paper develops an NFV-based behavior recognition system with low complexity to realize a bed-exit monitoring application on embedded systems. To achieve effectiveness and low complexity, a queueing-based behavior classification is proposed to keep memories of object tracking information and a specific behavior can be identified from continuous object movement. The experimental results show that the developed system can recognize three bed behaviors, namely off bed, on bed and return, for NFV images with accuracy rates of 95~100%.

The Lifelog Monitoring System for Honeybees: RFID and Camera Recordings in an Observation Hive

A typical honeybee colony contains more than 15,000 individuals, each with its own task related to supporting the hive and maintaining the colony. In previous studies on honeybees, observing individual animals’ behaviors has been a difficult and time-consuming task to understand the relationship between in-hive communication and environmental changes outside the hive, therefore it is necessary in any attempt to develop applying a remote sensing technology. To allow researchers to pass much of this tracking work on to computers, we have developed the lifelog monitoring system for honeybees, which uses RFID and Raspberry Pi camera recordings. Our preliminary experiments consisted of several tests aimed at identifying the optimal conditions for this system. First, two commercial RFID readers with antennas were compared in terms of their sensitivity to signals from RFID tags placed at various distances. We found that the UP16-1000-J2 reader was much more sensitive and had a longer effective range compared to the UP4-200-J2. The most sensitive region in the RFID antenna on the UP16-1000-J2 reader was 30 mm long and 5 mm wide at its center. Based on this preliminary information, we designed and built a passage from the interior of the observation hive to the outside so that all RFID-tagged bees could be detected individually by the RFID reader as they walked through the passage. Moreover, to detect the direction of either departure or arrival of each bee, we placed two RFID antennas under the passage between the observation hive and the outside, one near each end of the passage. All departure and arrival times of RFID-tagged bees were detected with their ID numbers. Using recorded data from these two RFID readers, we could measure how much time each tagged bee spent outside the hive. In addition to RFID recording on the passage, we also tracked all in-hive movements of numbered RFID-tagged honeybees. In-hive movements were simultaneously, comprehensively and automatically recorded via six Raspberry Pi camera modules arranged on the two sides of the observation hive. The cameras were set to record from 6:30 to 19:30 every day for one month, once or twice each year from 2015 to 2018. The in-hive behaviors of these bees were analyzed according to a simultaneous tracking algorithm that we developed for this purpose. Data from the monitoring system revealed that time spent outside the hive increased markedly after following the waggle dance. In addition to its findings on bee behavior, this study also confirms the effectiveness of our recording system combining RFID and Raspberry Pi cameras for honeybee lifelog monitoring.

Anxiety Reduction Through Close Communication with Robotic Media in Dementia Patients and Healthy Older Adults

The purpose of this paper is to provide an overview of the recent research on android robotic media with a focus on its effects on older adults and to present a discussion on the implications of the experimental results. Social isolation of older adults is a leading issue in healthcare. Patients with dementia experience symptoms, such as agitation, which can result in increasing care burden. Android robotic media have been shown to provide a feeling of safety and communication support to older adults. In previous case studies, an increase in prosocial behaviors was observed in participants with dementia; however, the media effects needed to be measured by assessment scales. The current results indicate the effects of robotic media on dementia symptoms, especially the decrease in patients’ anxiety. As demonstrated in another experiment, anxiety reduction can also be expected in healthy older adults; however, certain conditions may be required for both healthy older adults and those with dementia. Key factors for the media effect, namely, dementia type, user’s personality, and personalized dialogue are taken into consideration for the further development of robotic systems. Additionally, we further discuss the significance of long-term data collection, especially from the early life stages because teleoperation or autonomous systems are expected to utilize information that can affect the effect of robotic media.

Dangerous Situation Detection for Elderly Persons in Restrooms Using Center of Gravity and Ellipse Detection

We developed a restroom danger detection system (RDDS) for detecting dangerous situations and protecting the elderly. Restrooms are particularly dangerous places for elderly persons. Our RDDS detects danger in real time by using image processing and sends an alert to a family member, hospital staff, etc. It comprises four processes: person detection, center of gravity detection, ellipse detection, and danger decision. The human detection process calculates the difference between an image of the empty restroom and one of the restroom when it is occupied (to which a brightness correction has been applied). The difference image is binarized and used for detecting the presence of a person. If a person is detected, the person’s center of gravity and ellipse are detected in the binarized image after it is denoised. The obtained information is used for detecting a dangerous situation. If the dangerous situation continues for 60 seconds, an alert is sent. Testing showed that our system can detect a dangerous situation within 1.5 seconds. This RDDS is one step toward the development of a comprehensive elderly person protection system.

Application of Deep Learning to Develop a Safety Confirmation System for the Elderly in a Nursing Home

[abstFig src='/00290002/07.jpg' width='210' text='The motion detection system' ] Fall accident prevention is one of the most important issues in elderly care settings. To prevent an accident, it is necessary to notify caregivers if the elderly person is getting out of bed. We have previously developed a posture discrimination system based on body motions. Herein, we propose a discrimination method by using machine learning to improve the performance of the system. A purpose of this study is to evaluate the proposed method. Elderly people in a nursing home were chosen as subjects in this study. We analyzed the body motion data during bed rest and bed exit of the subjects using the proposed method. These results suggest that it is effective.

Screening Sleep Disordered Breathing with Noncontact Measurement in a Clinical Site

[abstFig src='/00290002/06.jpg' width='300' text='Respiratory rate from simulator and Kinect' ]<span class=”bold”>Background and purpose:</span>It has been considered that sleep-disordered breathing disorders, such as sleep apnea syndrome (SAS), cause an increase in the risk of cardiovascular disease or traffic accident risk, and thus early detection of SAS is important. It has been also important for medical workers at clinical sites to quantitatively evaluate the respiratory condition of hospitalized patients who are asleep in a simple method. A noncontact-type system was proposed to monitor the respiratory condition of sleeping patients and minimized patient-related stress such that medical workers could use the system for SAS screening and perform a preliminary check prior to definite diagnosis.<span class=”bold”>Method:</span>The system included Microsoft Kinect™ for windows® (Kinect), a tripod, and a PC. A depth sensor of Kinect was used to measure movement in the thorax motion. Data obtained from periodic waveforms were divided with the intervals of 1 min, and the number of peaks was used to obtain the respiratory rate. Additionally, a frequency analysis was performed to calculate the respiratory frequency from a frequency at which the maximum amplitude was observed. In Experiment 1), a METI-man® PatientSimulator (CAE healthcare) (simulator) was used to study the respiratory rate and frequency calculated from the Kinect data by gradually changing the designated respiratory rate. In Experiment 2), the respiratory condition of four sleeping subjects was monitored to calculate their respiratory rate and frequencies. Furthermore, a video camera was used to confirm periodic waveforms and spectrum features of body movements during sleep.<span class=”bold”>Results:</span>In Experiment 1), the results indicated that both the respiratory rate and frequency corresponded to the designated respiratory rate in each time zone. In Experiment 2), the results indicated that the respiratory rate of examines 1, 2, 3, and 4 corresponded to 12.79±2.44 times/min (average ± standard deviation), 16.46±4.33 times/min, 28.24±2.79 times/min, and 13.05±2.64 times/min, respectively. The findings also indicated that the frequency of examines 1, 2, 3, and 4 corresponded to 0.20±0.04 Hz, 0.26±0.06 Hz, 0.45±0.12 Hz, and 0.22±0.06 Hz, respectively. The periodic waveforms and amplitude spectra were enhanced with respect to body movements although regular waveform data were obtained after the body movement occurred.<span class=”bold”>Discussions:</span>The results indicated that body movement and posture temporarily affected monitoring of the system. However, the findings also revealed that it was possible to calculate the respiratory rate and frequency, and thus it was considered that the system was useful for monitoring the respiration confirm with the non-contact or SAS screening of patients in clinical site.