Research on IMU-Based Motion Attitude Acquisition and Motion Recognition

With the wide application of lower limb exoskeleton robots in various fields, problems such as individual motion feature recognition have gradually come to the fore. Therefore, this article carries out a study on motion pose acquisition and action recognition based on inertial measurement unit (IMU) sensors. By using a motion capture system to analyze the motion movements of the human lower limbs and establishing a human lower limb motion dataset, the motion laws of the lower limb joints were extracted. We also used IMU sensors for real-time motion data acquisition. The data obtained from the two acquisition methods were compared to ensure the reliability of the IMU sensor data acquisition. To improve the accuracy of data acquisition, the IMU sensor data were processed using Kalman filtering. Based on the dynamic time warping (DTW) algorithm, we carry out simulation experiments to analyze the motion data collected by IMU in real-time compared with the motion dataset, to achieve the real-time recognition of human motion movements. This research can solve the problem of individual motion feature recognition in the application of lower limb exoskeleton robots and provide strong support for research and application in rehabilitation and assisted walking.