Conference Papers

KAIST Exoskeleton Lab

2025 Real-Time Gait Mode Recognition of Multiple Stair-Climbing Modes for Hip-Assisting Wearable Robots

본문

Author
Chanyoung Ko, Kyoungchul Kong, Hyunjin Choi
Conference
10th IFAC Symposium on Mechatronic Systems & 14th IFAC Symposium on Robotics
Volume
59
Number
18
Page
223-228
Year of publication
2025

​Abstract:

This study proposes a real-time gait mode recognition algorithm for hip-assistive wearable robots during stair climbing, focusing on distinguishing step-over-step (SOS) and step-to-step (STS) modes. Biomechanical analysis of hip joint dynamics revealed key differences in swing phase kinematics and kinetics between the two modes. Two gait parameters—hip crossing ratio (ϕHC) and normalized angular velocity (θHC/norm)—were extracted and used to recognize gait modes using the DBSCAN clustering algorithm. The proposed system, implemented on a hip-assistive robot with embedded sensors, achieved a high recognition accuracy of 97.51% across subjects. These results demonstrate the feasibility and generalizability of recognition of multiple stair-climbing gait modes, enabling more adaptive and safe assistance strategies for wearable robotics.