Yutaka Deguchi

Multiple-robot monitoring system based on a service-oriented DBMS

In this paper, we present a human-targeted monitoring system composed of two autonomous mobile robots based on a service-oriented DBMS, mainly from the viewpoint of positioning control. Each robot is equipped with a Kinect and monitors the target human from appropriate angles and distances. The service-oriented DBMS, which manages the monitoring system and enables a rapid development and extension of the system, views each robot as a data source which generates a data stream to be stored and processed in the DBMS. The results of the experiments conducted in a real office are promising.

Skeleton Clustering by Autonomous Mobile Robots for Subtle Fall Risk Discovery

In this paper, we propose two new instability features, a data pre-processing method, and a new evaluation method for skeleton clustering by autonomous mobile robots for subtle fall risk discovery. We had proposed an autonomous mobile robot which clusters skeletons of a monitored person for distinct fall risk discovery and achieved promising results. A more natural setting posed us problems such as ambiguities in class labels and low discrimination power of our original instability features between safe/unsafe skeletons. We validate our three new proposals through evaluation by experiments.

Multi-view Onboard Clustering of Skeleton Data for Fall Risk Discovery

We propose a multi-view onboard clustering of skeleton data for fall risk discovery. Clustering by an autonomous mobile robot opens the possibility for monitoring older adults from the most appropriate positions, respecting their privacies, and adapting to various changes. Since the data that the robot observes is a data stream and communication network can be unreliable, the clustering method in this case should be onboard. Motivated by the rapid increase of older adults in number and the severe outcomes of their falls, we adopt Kinect equipped robots and focus on gait skeleton analysis for fall risk discovery. Our key contributions are new between-skeleton distance measures for risk discovery and two series of experiments with our onboard clustering. The experiments revealed several key findings for the method and the application as well as interesting outcomes such as clusters which consist of unexpected risky postures.

Developing a Face Monitoring Robot for a Desk Worker

We have developed an autonomous mobile robot which monitors the face of a desk worker. The robot uses three kinds of information observed with its Kinect to search for the desk worker and adjusts its position for monitoring. The monitoring is based on incremental clustering of the faces. Our experiments revealed that not only Animation Units (AUs) features, which represent deviations from the neutral face, but also the pitch angle of the face normalized in a new way are necessary for a valid clustering under specific conditions. Our robot lost sight of a desk worker only once in experiments for 8 persons for about 50 minutes. The resulting clusters correspond to “yawning”, “smiling”, and “reading” for a half of the desk workers with high NMI (normalized mutual information), which is an evaluation measure often used in clustering.

Skeleton clustering by multi-robot monitoring for fall risk discovery

This paper tackles the problem of discovering subtle fall risks using skeleton clustering by multi-robot monitoring. We aim to identify whether a gait has fall risks and obtain useful information in inspecting fall risks. We employ clustering of walking postures and propose a similarity of two datasets with respect to the clusters. When a gait has fall risks, the similarity between the gait which is being observed and a normal gait which was monitored in advance exhibits a low value. In subtle fall risk discovery, unsafe skeletons, postures in which fall risks appear slightly as instabilities, are similar to safe skeletons and this fact causes the difficulty in clustering. To circumvent this difficulty, we propose two instability features, the horizontal deviation of the upper and lower bodies and the curvature of the back, which are sensitive to instabilities and a data preprocessing method which increases the ability to discriminate safe and unsafe skeletons. To evaluate our method, we prepare seven kinds of gait datasets of four persons. To identify whether a gait has fall risks, the first and second experiments use normal gait datasets of the same person and another person, respectively. The third experiments consider that how many skeletons are necessary to identify whether a gait has fall risks and then we inspect the obtained clusters. In clustering more than 500 skeletons, the combination of the proposed features and our preprocessing method discriminates gaits with fall risks and without fall risks and gathers unsafe skeletons into a few clusters.

Toward a platform for collecting, mining, and utilizing behavior data for detecting students with depression risks

In this paper, we present our plan for constructing a platform for collecting, mining, and utilizing behavior data for detecting students with depression risks. Unipolar depression makes a large contribution to the burden of disease, being at the first place in middle- and high-income countries. We survey descriptors of depressions and then design a data collection platform in a classroom based on the assumption that such descriptors are also effective to students with depression risks. Visual, acoustic, and e-learning data are chosen for collection and various issues including devices, preprocessing, and consent agreements are investigated. We also show two kinds of utilization scenarios of the collected data and introduce several techniques and methods we developed for feature extraction and early detection.

Hidden Fatigue Detection for a Desk Worker Using Clustering of Successive Tasks

To detect fatigue of a desk worker, this paper focuses on fatigue hidden in smiling and neutral faces and employs a periodic short time monitoring setting. In contrast to continual monitoring, the setting assumes that each short-time monitoring (in this paper, it is called a task) is conducted only during a break time. However, there are two problems: the small number of data in each task and the increasing number of tasks. To detect fatigue, the authors propose a method which is a combination of multi-task learning, clustering and anomaly detection. For the first problem, the authors employ multi-task learning which builds a specific classifier to each task efficiently by using information shared among tasks. Since clustering gathers similar tasks into a cluster, it mitigates the influence of the second problem. Experiments show that the proposed method exhibits a high performance in a long-term monitoring.

Continuous query processing over data, streams and services: Application to robotics

Developing applications involving mobile robots is a difficult task which requires technical skills spanning different areas of expertise, mainly computer science, robotics and electronics. In this paper, we propose a SQL-like declarative approach based on data management techniques. The basic idea is to see a multi-robot environment as a set of data, streams and services which can be described at a high level of abstraction. A continuous query processing engine is used in order to optimize data acquisition and data consumption. We propose different scenarios to classify the difficulty of such an integration and a principled approach to deal with the development of multi-robot applications. We provide our first results using a SQL-like language showing that such applications can be devised easily with a few continuous queries.