Upper Body Posture Recognition Using Inertial Sensors and Recurrent Neural Networks

Abstract

Inadequate sitting posture can cause imbalanced loading on the spine and result in abnormal spinal pressure, which serves as the main risk factor contributing to irreversible and chronic spinal deformity. Therefore, sitting posture recognition is important for understanding people’s sitting behaviors and for correcting inadequate postures. Recently, wearable devices embedded with microelectromechanical systems (MEMs) sensors, such as inertial measurement units (IMUs), have received increased attention in human activity recognition. In this study, a wearable device embedded with IMUs and a machine learning algorithm were developed to classify seven static sitting postures: upright, slump, lean, right and left bending, and right and left twisting. Four 9-axis IMUs were uniformly distributed between thoracic and lumbar regions (T1-L5) and aligned on a sagittal plane to acquire kinematic information about subjects’ backs during static-dynamic alternating motions. Time-domain features served as inputs to a signal-based classification model that was developed using long short-term memory-based recurrent neural network (LSTM-RNN) architecture, and the model’s classification performance was used to evaluate the relevance between sensor signals and sitting postures. Overall results from performance evaluation tests indicate that this IMU-based measurement and LSTM-RNN structural scheme was appropriate for sitting posture recognition.

Intelligent systems for sitting posture monitoring and anomaly detection: an overview

“…4 The image on the left shows the location of the inertial sensors between the C7 and L4 vertebrae provided by [ 49 ]. On the right is an example of implementation of the sensors on the spine of a test subject of the [ 45 ] work …”

“…Among the different existing wearable sensors, inertial sensors or IMUs are the most widespread for postural monitoring [ 16 , 45 , 47 , 49 ]. It is a sensor composed of an accelerometer, a gyroscope and a magnetometer, allowing to know the angular velocities and linear accelerations of those points of the body where it is located, being able to perform a study of the movement.…”

Double-Fishtail-Shaped FBG Wearable Device for Sitting Posture Recognition and Real-Time Respiratory Monitoring

“…Its retroverted curve allows the spine to resist tension and protect the spinal cord when bending or moving the body. Due to the thoracic spine providing most of the support and stability for the entire trunk, poor posture-induced spinal deformities often occur in this segment [12]. Consequently, the sensor devices were placed in the thoracic spine region.…”

“…Instead, wearable devices [8,9] are a viable alternative due to their weak effect on the worker's posture and allow for constant monitoring. Traditionally, goniometers [10], x-rays [11], and motion capture systems (Motion Capture Unity) [12,13] are commonly used devices for detecting UBMDs in healthcare practice. While accurate, these devices have proven to be troublesome or uncomfortable to use throughout the workday.…”

A Low-Cost Inertial Measurement Unit Motion Capture System for Operation Posture Collection and Recognition

“…A comparison of prices and performance of other IMU solutions is presented in Table 1 , covering aspects such as price, sampling rate, accelerometer rate noise spectral density, gyroscope rate noise spectral density, interface mode, and battery life. The IMU solutions commonly used in human motion recognition research, such as Xsens MTw Awinda [ 10 ], MetaMotionR [ 10 ], Next-Generation IMU [ 11 , 12 ], MetaMotionC [ 13 ], Shimmer3 [ 14 ], and InvenSense MPU-9250 [ 15 ], were chosen. Through comparison, our solution demonstrates advantages in terms of pricing.…”