Wearable Continuous Gait Phase Estimation During Walking, Running, Turning, Stairs, and over Uneven Terrain
Linghui Xu,Yuting Chen,Bingfei Fan,Canjun Yang,Wei Yang
DOI: https://doi.org/10.1109/tmrb.2024.3407366
2024-01-01
IEEE Transactions on Medical Robotics and Bionics
Abstract:Wearable continuous gait phase estimation is essential for walking assistance, clinical rehabilitation, and clinical assessment; however, most algorithms have only been validated for straight-line and constant-speed walking, and it is unclear how performance will change in real-life locomotion scenarios. A generalized paradigm is needed to comprehensively assess and recommend wearable continuous gait phase estimation strategies for the diverse array of walking situations. We thus propose a comprehensive evaluation indicator system for eight typical gait activities in daily life including slow walking, standard walking, running, walking with turns, stair descent, stair ascent, stop-and-go, and uneven terrain walking. The indicator system was used to evaluate four commonly used continuous gait phase estimation strategies: adaptive oscillators, phase oscillator, neural network, and time-based estimation. Eleven healthy participants were enrolled in the evaluation. All estimation strategies performed well for constant-speed walking but performance varied for other activities. Time-based estimation was most accurate for slow walking (0.094±0.011 rad root mean square error, 1.50±0.18% of one gait cycle), running (0.167±0.028 rad, 2.66±0.44%) and walking with turns (0.124±0.047 rad, 2.00±0.75%). Adaptive oscillators were most accurate for standard walking (0.115±0.037 rad, 1.83±0.59%). Phase oscillator was most accurate for stair climbing (0.280±0.063 rad, 4.46±1.00%) and uneven terrain (0.204±0.069 rad, 4.30±1.10%). Neural network was most accurate for stop-and-go (0.27±0.114 rad, 4.30±1.81%). These results can potentially provide guidance for determining suitable gait phase estimation strategies in realistic locomotion scenarios, and in comparing and optimizing the current proposed strategies.