A joint research and development team consisting of Kohtaro Ohba, Group Leader of the Ubiquitous Functions Research Group of the Intelligent Systems Research Institute (Shigeoki Hirai, Director) of the National Institute of Advanced Industrial Science and Technology (AIST; Hiroyuki Yoshikawa, President), and Masateru Minami, Assistant Professor of the Laboratory for Information Systems and Applications of the Faculty of Engineering of the Shibaura Institute of Technology (Takao Nagatomo, Chief Director) have combined wireless network nodes consisting of over 100 temperature, humidity and other sensors, with robotic middleware technology (RT middleware) to form a highly effective system for collecting information.

The materialization of a monitoring system for routine maintenance and management of buildings, bridges, plants, and other structures and facilities, requires large-scale sensors and a network to connect them. However, as the scale of the facility increases, the scale of the sensor network also increases, and costs rise accordingly. In response, the AIST has developed a highly reliable, low-cost large-scale sensor network by attaching sensors to previously developed super energy-saving wireless network nodes.

However, middleware technology is required to improve the development effectiveness of the widely scattered sensors, and to provide an easy-to-use interface for users. For this purpose, the use of RT middleware developed at the AIST indicated that a highly effective system could be constructed.

The system developed in this project was exhibited at the Open House of Intelligent Systems Research Institute, AIST scheduled for November 15, 2006.

 

Figure 1. Examples of a large-scale monitoring system

Since the launch of the MOTE Project in USA, much research has been conducted on sensor networks. Since the business market for this technology is expected to be big, there is great demand to produce a marketable product as soon as possible. However, the sensor network nodes that are currently sold commercially are too functional, expensive, and impractical because they are designed for research applications. In order to monitor large structures such as buildings, bridges and plants, it is necessary to connect widely scattered sensors easily and cheaply.

Much interdisciplinary research has already been conducted on such sensor networks, but this has been confined to laboratories, and experiments have mainly involved the use of equipment consisting of 20-30 very expensive (20,000 yen or more each) sensor network nodes; there has been no research on developing sensor network nodes that can be attached to large-scale systems such as buildings, bridges and plants. Moreover, there has been no actual operation of a system containing 100 or more nodes, so industry has had doubts about the practicability of such a system.

The Intelligent System Research Institute of the AIST has been researching methods of controlling environmentally embedded robots that contain scattered robot elements/components by utilizing wireless systems, networks, etc., that are found in the human living environment.

Middleware technology linking robot components was developed to enhance modularity and reusability when writing robot control programs, and work has been proceeding on formulating international standards. In a project being undertaken by the New Energy and Industrial Technology Development Organization (NEDO), middleware technology is currently being provided as Open RTM. Research has also been proceeding on developing wireless network nodes, etc., that can be used for wirelessly connecting robot components with one another.

Utilizing the potential of this research, work started on developing a practical sensor network. This research and development is a part of the FY2006 R&D Project for Supporting Small and Medium Companies in Local Communities.

The sensor network developed in the present project consists of more than 100 temperature, humidity and other sensors as nodes in a wireless sensor network, and a system that can read temperature, humidity, and other data using RT middleware. The temperature and humidity sensors, which were developed using MEMS technology, contain an AD converter that can transmit digital information to network nodes. This helps to greatly suppress noise that normally exists in analog parts and increase the accuracy of the information.

The ad hoc communications used by most sensor networks contain network nodes that require CPU or other power. This makes the network nodes to expensive to use, and the network structure is generally a star-shaped communication system. However, the present project has created network nodes that are inexpensive and can be used to build simple, practical network systems.

The previously developed wireless network nodes enable highly reliable wireless communications even in environments containing large amounts of metal or electromagnetic waves. Moreover, since very little electric power is consumed, environmental sensors can be run by battery power. These network nodes are reliable and inexpensive (only about 1/10 the price of conventional nodes), and are already being received by the market.

So far, there have been numerous examples of sensor networks and ubiquitous devices such as IC tags being used to construct “intelligent spaces” in living environments. However, there have been few examples of constructing low-cost, easy-to-install high-reliability sensor networks that can be used for industrial applications such as in the present project. We are approaching the day when wireless and RT middleware technologies will help to bring about sensor network systems with dispersed installations that can be run by battery power.

By developing wireless network nodes and RT middleware technology as robot element technology, it may become possible to utilize robot elements/components that have hitherto been prohibited by the high costs associated with wiring, program development, etc., and even develop a robot system industry. Future plans call for achieving this goal through the application and verification of various system architecture technologies as sensor networks.

Figure 2. 100 of the Sensor network nodes

Figure 3. Sensor implementation image