Columns that reveal the world
- Getting up close and personal with the researchers -

I am researching what can be achieved by using sensors.
A sensor is a device or equipment that detects changes in the physical state of a phenomenon or object, such as the magnitude of force, distance, or brightness, and converts it into a signal or data for output. Many people may find this definition difficult to understand.

It is true that conventional sensors were mainly built into factory machines for control, labor saving and quality control, or built into devices and products to enable automation, ease of use and improved safety. All of these were key technologies that made a significant contribution to strengthening the competitiveness of Japanese industrial products after the war, but their functions were extremely limited and we did not feel their existence in our daily lives.

However, in recent years, sensors have rapidly become a part of everyday life thanks to the spread of smartphones. These small devices, of which nearly 600 million are shipped worldwide each year, are actually packed with a variety of sensors.

For example, when you tilt your smartphone sideways, the screen also rotates sideways in response to that movement. This is due to the action of the gyro sensor. A gyro sensor is a sensor that measures angular velocity (how much the angle of an object changes per unit time, in other words, the speed at which the object is rotating).

Unlike mechanical switches that turn the current on and off through physical contact, proximity sensors are sensors that can switch the current on and off simply by having an object approach. In smartphones, this is useful for functions such as automatically turning off the touch panel display when you hold the device close to your ear during a call, preventing malfunctions.

In addition, many other sensors are housed in that small device, including a brightness sensor, an acceleration sensor, a gravity sensor, an air pressure sensor, and a temperature sensor. And we enjoy the functions of these various sensors in the form of "apps."

Thus, we are now in an era where smartphones are equipped with a wide variety of sensors that would have been unimaginable just a generation ago, and people carry them around with them.

With the rapid spread of smartphones, research and development into sensors with even higher functionality and performance is progressing at a rapid pace. Sensor utilization technology holds great potential.

The spread of smartphones and wearable devices has created a situation in which "many people carry high-performance sensors with them at all times" and at the same time, "high-performance sensors are constantly connected to a network." This "sensor network" is a mechanism that provides solutions to various problems by connecting sensors attached to all kinds of devices in society, social infrastructure (buildings, roads, railways, etc.), and people (smartphones, wearable devices, etc.) to a network and collecting and analyzing huge amounts of data from them.

For example, one weather information service company uses reports and location information from members' smartphones, as well as images of the sky taken with cameras, to forecast the weather. They use the "overwhelming power of numbers" of smartphones to collect data that cannot be captured by their own observation network alone.

Other systems under consideration include installing sensors on bridges, tunnels, etc. for constant monitoring, which would issue warnings calling for inspections if cracks or other abnormalities occur, and systems that would allow the extent of damage to infrastructure in the event of a disaster to be grasped in real time without the need for humans to go to the site.

Furthermore, sensor networks are also useful in addressing issues such as rising medical costs and an aging population. The Japanese government is promoting home medical care in order to reduce medical benefit costs. In this context, wireless sensing is positioned as a means of supporting health management. The basic idea is to use wearable sensors attached to the human body to collect information such as body temperature and heart rate. A Japanese manufacturer has commercialized a wristband-type device equipped with a heart rate sensor and an acceleration sensor for the purpose of health management. By using a heart rate sensor and an acceleration sensor, it is possible to accumulate daily life rhythms and physical condition such as sleep, walking, and indoor walking as time-series data. By comparing current data with past data, if any abnormal changes are observed, the system can determine that this is an "abnormal change in physical condition" and take action such as reporting to a medical institution immediately.

In this way, it is hoped that sensor networks will be used to solve various social issues and create new businesses.

In this context, the theme that our research group is currently working on is "sensor technology that helps facilitate smooth communication between people."

Take meetings, for example. Meetings are an important way to make decisions and reach consensus in a variety of groups, whether at work, school, home, or in the local community, but many people find it difficult to get all participants to speak up and quickly gather their opinions.

In order to hold lively meetings, we need an objective indicator to evaluate the "liveliness of the meeting." If we had such an indicator, we could compare and analyze the liveliness of different meetings and provide advice to encourage lively meetings. Therefore, in our laboratory, we are researching a system to define an indicator of meeting liveliness and judge whether a meeting is good or bad.

In this research, audio information from meetings is sensed and analysis is conducted to determine which participants are speaking, after which the activity level of the meeting is quantified using the following three indicators:

(1) Fairness of discussion: An active meeting requires that all participants participate in the discussion fairly. This is an indicator to measure whether everyone is speaking equally by paying attention to the speaking time of each member.

(2) Discussion dominance: In order to encourage lively discussion, it is important to have a leader who acts as a moderator. Here, we focus on the frequency with which each member speaks, and consider the member who speaks most frequently to be the leader.

(3) Discussion Mediation Level: In order to draw out the opinions of members, it is necessary for the leader to give equal rights to speak. Here, we measure whether the leader gives equal rights to speak by analyzing "who speaks after the leader speaks."

"KAIHUI" is a system that analyzes the above information to find an index of meeting activity. In the laboratory, we used KAIHUI to measure meeting activity in six types of meetings that differed in "gender of participants," "meeting topic," and "way of proceeding with discussion."

This research has only just begun. In the future, it will be necessary to create indicators that focus on the "quality" of meetings by adding elements such as the content of participants' comments, facial expressions, and gestures. By continuing this research, we may one day be able to create a "facilitator robot." This robot will encourage participants to speak, sometimes with witty jokes, organize the flow of the discussion, confirm that participants' perceptions are consistent, and support consensus building and mutual understanding.

Sensor technology has advanced by making it possible to measure things that were previously unmeasurable. With future research, it may become possible to measure things like happiness and comfort. The development of sensor technology holds infinite possibilities.

(Published in 2016)