Leg-Joint Angle Estimation from a Single Inertial Sensor Attached to Various Lower-Body Links during Walking Motion

Gait analysis is important in a variety of applications such as animation, healthcare, and virtual reality. So far, high-cost experimental setups employing special cameras, markers, multiple wearable sensors have been used for indoor human pose-tracking gait-analysis purposes. Since locomotive activities walking are rhythmic exhibit kinematically constrained motion, fewer can be employed gait pose analysis. One the core parts lower-limb-joint angle estimation. Therefore, this study proposes neural network-based angle-estimation method from single inertial sensor unit. As proof concept, four different neural-network models were investigated, including bidirectional long short-term memory (BLSTM), convolutional network, wavelet unidirectional LSTM. Not only could selected network affect estimation results, but also placement. Hence, waist, thigh, shank, foot candidate positions. From these sensors, two sets angles estimated. set contains sagittal-plane leg-joint angles, while second includes six coronal-plane angles. After assessment combinations networks datasets, BLSTM with either shank or thigh datasets performed well both joint-angle sets. better candidates sensor-based Consequently, mean absolute error (MAE) 3.65° 5.32° four-joint-angle eight-joint-angle obtained, respectively. Additionally, actual leg motion was compared to computer-generated simulation predicted joints, which proved possibility estimating during