An Intelligent Head Posture Recognition Pillow

The position and posture of the head during sleep can reflect the quality of sleep and overall health. As an important component of bedding that supports the head during sleep, pillows have gained attention from researchers due to their potential to monitor head posture through a variety of sensing means. This paper proposes a non-wearable sleep posture recognition system based on microphones and speakers integrated into the pillow. Two speakers and four microphones are utilized as sound sources and receivers to identify the position and posture of the human head on the pillow. To monitor posture, the pillow is divided into five areas to detect positions related to supine and lateral positions. Due to the pressure exerted by the head on the pillow, the inner latex material of the pillow will be deformed. At the same time, the microphones will be squeezed, changing the characteristics of the sound received. After analyzing the features of the collected data, fully connected neural network is used for classification. The position and posture of the human head on the pillow corresponding to the sound fragment can be predicted by the similarity measurement calculated from the features. The method achieves a posture recognition accuracy of 91.1% and a position recognition accuracy of 98.3%. In conclusion, this study creatively demonstrates the feasibility of identifying the position and posture of the human head on the pillow using speakers integrated into the pillow as a sound source. This method can be combined with hypnotic music in practical applications, making it highly promising in relieving insomnia and improving sleep quality.