Search

Agile environmental monitoring is a novel style of environmental-monitoring development that lets you rapidly develop sensor devices and monitoring systems through trial and error during continuous sensing. The authors developed three environmental-monitoring systems and adapted them in situ based on actual field requirements. This article is part of a special issue on smart cities.

Thanks to the prevalence of mobile phones and GPS devices, spatiotemporal population data can be obtained easily. In this article, we propose a mixture of collective graphical models for estimating people flow from spatiotemporal population data. The spatiotemporal population data we use as input is the number of people in each grid cell area over time, which is aggregated information about many individuals; to preserve privacy, they do not contain trajectories of each individual. Therefore, it is impossible to directly estimate people flow. To overcome this problem, the proposed model assumes that transition populations are hidden variables and estimates the hidden transition populations and transition probabilities simultaneously. The proposed model can handle changes of people flow over time by segmenting time-of-day points into multiple clusters, where different clusters have different flow patterns. We develop an efficient variational Bayesian inference procedure for the collective graphical mixture model. In our experiments, the effectiveness of the proposed method is demonstrated by using four real-world spatiotemporal population datasets in Tokyo, Osaka, Nagoya, and Beijing.

A concentration of people attracted to urban areas or large events raises demand for crowd traffic control systems. Such systems need to achieve a collective control for dealing with the crowd interacting with each other, as well as a flexible control for managing the traffic depending on application scenarios such as swift passage on streets or safety-guaranteed traffic. This paper presents a Bayesian optimization (BO) approach using a multi-agent simulation for efficiently finding a control of traffic: the multi-agent simulation considers the interaction of people changing their routes for a collective control, whereas the black-box nature of BO is capable of replacing the optimization metric for different scenarios without algorithmic modification to achieve flexible traffic management. Experiments use two maps: synthetic grid-like streets and the Enoshima Coast firework festival site in Japan. In both cases, the results demonstrate that our traffic control is effective for optimizing various metrics such as an efficiency-oriented arrival rate, a safety-oriented low traffic rate, and their combination.

We present how crowdsensing with civil officers contribute to collect larger amount of urban data with high spatial-temporal coverage, and enable to understand urban features by analyzing the data. We implemented a crowdsensing system which fits to daily city management, and conducted long-term experiment with garbage section of Fujisawa city, Japan. Based on analysis of collected data, we show that our system provide fine-grained urban data compared to crowdsensing with citizens with improving efficiency of city management. In addition, we analyze our dataset with demographic urban data, and clarified the relationship between pattern of uncollectible garbage thrown and features of resident area.

Garbage is intricately with our daily life. We are investigating method that estimates regional amounts of garbage using motion sensors mounted on garbage trucks. In this paper, we report out analysis results of garbage amounts using national census data. We could obtain insightful information from just motion sensors we mounted on garbage trucks.

Machine learning is a promising technology for analyzing diverse types of big data. The Internet of Things era will feature the collection of real-world information linked to time and space (location) from all sorts of sensors. In this paper, we discuss spatio-temporal multidimensional collective data analysis to create innovative services from such spatio-temporal data and describe the core technologies for the analysis. We describe core technologies about smart data collection and spatio-temporal data analysis and prediction as well as a novel approach for real-time, proactive navigation in crowded environments such as event spaces and urban areas. Our challenge is to develop a real-time navigation system that enables movements of entire groups to be efficiently guided without causing congestion by making near-future predictions of people flow. We show the effectiveness of our navigation approach by computer simulation using artificial people-flow data.

The datafication of spatio-temporal city-wide events is one essential factor for smart management of the city. For this purpose, the combination of real-time event detection on edge sensor nodes mounted public vehicles, and event accumulation on a server is one realistic and efficient solution. We can analyze the accumulated data to understand complex phenomena occurring in entire the city. In this paper, we introduce a novel datafication procedure of city-wide events by sensor mounted garbage trucks and evaluated the preliminary implementation of event detection system on actual vehicle-mounted sensors.

In this paper, we propose a method for directly measuring and analyzing driving behavior using wireless 3D-accelerometers. Whereas existing systems installed many sensors into a specially equipped test vehicle to indirectly measure driving behaviors, our method uses wireless 3D-accelerometers attached to a driver for directly measuring his/her behaviors in a vehicle. After applying independent component analysis (ICA) to reduce car-caused noise, our method detects anomalies in driving behaviors using one-class SVM. By directly measuring driving behavior, our method allows to point out anomalies in driving behaviors characteristics to novice drivers with precision of 71.1% and recall of 73.9%.

This paper proposes a new method that makes it easy for us to construct a positioning model for device-free passive indoor localization by using model transfer techniques. With device-free passive indoor positioning, a wireless sensor network is used to detect the movement of a person based on the fact that RF signals transmitted between a transmitter and a receiver are affected by human movement. However, because device-free passive indoor positioning relies on machine learning techniques, we must collect labeled training data at many training points in an end user's environment. This paper proposes a method that transfers a signal strength model used for locating a person obtained in another environment (source environment) to the end user environment. With the transferred models, we can construct a positioning model for the end user environment inexpensively. Our evaluation showed that our method achieved almost the same positioning performance as a supervised method that requires labeled training data obtained in an end user's environment.

This paper compares the effect of a robot's and on-screen agent's recommendations on human decision-making using a quantitative evaluation method. We are interested in whether a robot's physical body produces some differences in the effect or not. Previous research investigated the advantage of a physical body; however, the advantage was not clarified quantitatively and there was not enough evidence to give the results credibility. Our method based on quantitative evaluation clarifies the effect of a robot's and on-screen agent's behavior on user decision-making. Comparing a robot's behavior with an on-screen agent's, we show that the degree of the effect firmly depends on the interaction environment and that geometrical consistency between the interaction environment and embodied social agents (ESAs), which include robots and on-screen agents, is important in the recommendation situation.

We present a novel approach for real-time and proactive navigation in crowded environments such as event spaces and urban areas where many people are moving to their destinations simultaneously. Our challenge is to develop a real-time navigation system that enables movements of entire groups to be efficiently guided without causing congestion by making near-future predictions of people flow. Our approach tries to detect future congestion by using a spatio-temporal statistical method that predicts people flow. When future congestion is detected, our approach creates an optimal navigation plan based on "what-if" simulations, which accounts for the effect of total people flow change caused by navigation. We experimentally compare the spatio-temporal statistical method with the conventional matrix factorization based approach using a real data set. We also demonstrate the effectiveness of our navigation approach by computer simulation using artificial people-flow data.

Conservation of biodiversity is an important issue. We are investigating a system for monitoring the habitat of endangered fish (Japanese rosy bitterling) using a wireless sensor network. Accordingly, measurements are taken for dissolved oxygen (DO), water temperature, air temperature, humidity, and illuminance. In this paper, we describe this habitantmonitoring system for the rosy bitterling.

We introduce a new data compression method for efficient data gathering in hierarchical sensor networks. Our proposed method compresses sensor data sequences by decomposing them into local patterns and weight variables using Singular Value Decomposition (SVD). Our proposed method can achieve efficient data gathering for environmental monitoring.

We have constructed a sensor network in a real hospital environment to develop a system that prevents medical accidents by monitoring nursing activities. The network has been carefully designed to be dependable for practical purposes even in a network-unready environment with considerations of safety and stability as well as the consistency of the obtained sensor data. This paper describes the design and implementation of our sensor network. Through an experiment of about one week, we confirmed that our design performs well in a practical environment and obtains consistent data among the sensors worn by nurses and installed in the environment. Since a hospital has strict limitations on the use of sensor network equipment, the design described in this paper provides a practical solution for the construction of sensor networks in many indoor applications.

We propose a method for recognizing workers' routine activities by combining location information and body movements of a user. We describe methods for capturing and recognizing the nursing context with an infrared-ID location sensor system and accelerometers worn by a user. We show experimental results of recognizing typical activities in a nursing scenario in a laboratory setting.

This paper presents a wearable sensing device called "B-Pack", which is designed to capture nursing activities in real hospitals. To provide usability and flexibility, B-Pack features wireless connectivity using Bluetooth and a rechargeable battery with on-board charging circuitry. Furthermore, to acquire synchronized data on distributed parts of wearer's body, B-Pack features a time synchronization protocol. Through experiments and interviews, we confirmed that the data from each body part was appropriately collected for nursing activity recognition and that the design is suitable for a real nursing environment.

This paper proposes a Bluetooth-based indoor proximity detection method for nursing context awareness. We exploit proximity between Bluetooth devices attached to people and objects for estimating 1) room-level proximity of people and objects, and 2) mutual proximity between moving people and objects. We show that the proximity information exchanged between several devices can be updated at a rate of more than 1 Hz by effectively choosing the timing parameters of Bluetooth inquiry mechanism. Empirical results of evaluating the receiver signal strength indicator (RSSI) at various distances between devices are also shown.

This paper proposes another dynamic prediction procedure that observes nurses' conversations to determine critical points in nursing accidents or incidents. The points would be be suggested by words that have a certain feature. We regard such words that might cause accidents or incidents as chances. This paper analyzes these types of words and describes their features to dynamically determine the point at which an accident or incident occurs.

This paper describes a type of multiagent-based distributed manipulator control in which the execution of a task involves the interaction between multiple software modules called agents. The intent of this work is to integrate a model-based approach and a behavior-based approach under a multiagent control paradigm. This paradigm is applied to a manipulator control problem without the use of inverse kinematics. A conventional manipulator is thought of as a system composed of a group of spatially distributed freedoms, each of which is independently controllable. Our manipulator control system has multiple functional agents each of which individually attempts to contribute in a positive way toward improving the degree of the given task achievement. The experimental system for multiagent-based distributed control includes a Lisp machine, two types of sensors (force/vision) and a manipulator. An attempt is made to evaluate our control paradigm by giving the system an example task that involves the end-effector movement between two points. Agents are implemented as multiple concurrent processes programmed in Lisp.

This paper presents a robotic control architecture and descriptions based on interactions between multiple software modules called agents. When we apply this agent architecture, having modularity and flexibility, to a distributed robotic system, its sufficient definition and description methods are required because each agent communicates with the others by using its own functions directly or indirectly, recognizing or referring to its own slate and those of the others. We first assigned classes to the basic agent architecture of the multiagent system. This system was constructed with object-oriented programming, which consists of data (an internal state, external information and a behavior strategy) and methods. We then defined a finite state machine (FSM) as the model that the agent refers to in taking actions corresponding to its state transition. We applied this description and the model to controlling a robotic manipulator with multiple joints. The FSM model of the agent reflects both its internal state and its joint movements, i.e., forward, backward and stop. To perform a system evaluation of this approach, we have been creating an agent development environment and describing agents that will carry out an approaching task.

We developed a wireless sensor network system for supporting context-awareness of nursing activities in hospitals. Our system is aimed at automated recording of nursing work, providing context-aware services to nurses and visualising analytical results from nursing histories. The system consists of heterogeneous devices utilising three kinds of wireless networks: a ZigBee TM (ZigBee is a registered trademark of the ZigBee Alliance)-based location sensor network, a Bluetooth TM (Bluetooth is a trademark owned by Bluetooth SIG, Inc., USA)-based body-area sensor network for capturing nurse activities and a Wi-Fi TM (Wi-Fi is a registered trademark of the Wi-Fi Alliance)-based network for communication between PDAs worn by nurses and a server PC. We illustrate the layered structure of the entire system as well as algorithms for estimating nurse locations and activities during their operations in a hospital. These estimated results are managed by a real-time database system with which nurse contexts can be visualised in real time on a server PC and/or PDAs. We also show empirical results evaluating the performance of retrieving, analysing and visualising detailed nursing contexts.

This paper presents the preliminary results of classifying human touching behaviors using a haptic interface for a pet-like robot. The haptic interface uses gridded pressure-sensitive conductive ink sheets. Features of the measured pressure data are determined for classification in terms of 1) absolute values, 2) spatial distributions and 3) the temporal differences in measured pressure patterns. Touching behaviors include "slap," "pat," "stroke" and so forth. The experimental results show that a reliable classification of these touching patterns can be accomplished by using the sensor sheet and pressure features. The results of classification can be used as reward signals for reinforcement learning in controlling the behaviors of a pet-like robot that interacts with humans.