This research builds on the probabilistic approach for recognizing Sit-to-Stand (Sit2St) and Stand-to-Sit (St2Sit) transitions using a single Inertial Measurement Unit (IMU) sensor [1]. The primary objective is to develop an intelligent, real-time recognition system capable of identifying sitting, standing, and transitional movements with high accuracy in younger and older adults. The goal is to leverage this capability to enhance users’ experience with Sit2St wearable assistive technologies.
A Bayesian classifier was trained on filtered IMU data and tested on raw signals to ensure generalizability and resilience to noise [1]. An angular threshold approach was employed to segment the transition regions of Sit2St and St2Sit into three transition phases. The system achieved 100% accuracy in recognizing both activities and phases using only one sensor. The classifier was deployed on a microcontroller, demonstrating standalone operation and suitability for exoskeletons. Future steps of this research study involve exploring biological indicators of transition phase breakpoints and integrating the algorithm into an exoskeleton while maintaining semi-passivity.
In addition to algorithm development, the study explored age-related variations in Sit2St and St2Sit transitions across different age groups ranging from 18 to 95 years old. Metrics such as the duration of Sit2St and St2Sit transitions, acceleration, acceleration standard deviation, and frequency content were analyzed, revealing that transition duration may be more informative of mobility status than age alone. The goal of this study is to determine the quality of the subject’s mobility and to quickly identify whether assistance is needed.

Reference
[1]
Martinez-Hernandez, U., and Dehghani-Sanij, A. A., 2019, “Probabilistic Identification of Sit-to-Stand and Stand-to-Sit with a Wearable Sensor,” Pattern Recognition Letters, 118, pp. 32–41. https://doi.org/10.1016/j.patrec.2018.03.020.

Bridging technology and human motion to enhance lives across ages.