Possibility of wireless sensing leaded by Origin Wireless made by Wi-Fi/5g radio waves

In the 2020s, when IoT/AI technology has become widespread and the social implementation of cyber fidical systems (CPS) progresses, the “Sensing era” in which sensors that digitize people and things are embedded in every corner of society. Maybe. However, building a sensor network, which is the basis, is not as easy as it is. It is costly to install a large amount of physical sensors such as infrared sensors, etc., supply power, connect to the network, and continue to operate. If the number becomes enormous, for example, the maintenance of the sensor shifts or the battery is cut off, for example. If you are a surveillance camera, you can shoot a wide range of images with one unit and analyze it with AI, but this is a revenge on "too visible". There is a great risk of infringing privacy. The promising solution that solves these issues is "Wi-Fi Sensing", a technology that detects people and things by radio waves. ORIGIN Wireless, a venture from Maryland University that leads this field, "People, Pets, Rumba?" "I understand that people are moving, invading, falling down, and breathing even more," said Masato Marumo, the representative director of a Japanese corporation.

Origin Wireless JAPAN Representative Director Masato Marumo

同社の独自技術「WirelessAI」は、Wi-Fi電波の反射によって生じる“乱れ”を捉え、AIで変化の起因を特定するという手法を採用する。反射が生じる屋内であれば、送信機と受信機を設置するだけで、ある程度広い空間内の人の動きやドアの開閉などを検知することが可能だ。駅のプラットフォームやビル谷間のような半屋内でも使用可能なことを実験で実証している。同技術の注目すべき点はもう1つある。ディープラーニングによって検知・判定精度を高められることだ。検知データをAIに学習させ、その空間特有の変化に合わせて検知精度を高めていくことで、使えば使うほど“わかること”が増えていく。例えば、「検知した動きが呼吸による胸の鼓動よりも小さければ、ペットではないか、あるいはルンバじゃないかといったことが徐々に特定できるようになってきている」（丸茂氏）。