Balance Map Analysis for Visualization and Quantification of Balance in Human Walking

Evaluation of stability or loss of balance in walking persists as an open question. Although an inverted pendulum model is often adopted to evaluate stance leg balance, a stumbling-related balance loss should be associated with the swing leg. We propose a new framework based on a compass gait model that determines whether the current state (i.e., position and velocity) in the swing phase can maintain steady state walking or, instead, fall without active joint torque, which is termed as balance map analysis. The forward and backward balance loss regions are derived by a linear compass gait model. To test the balance map analysis, measurement experiments of steady state walking and stumbled walking are used to validate two hypotheses: 1) the state during steady-state walking is not located in the balance loss region; and 2) if stumbling occurs, the state moves toward the forward balance loss region. The results of the balance map analysis showed good agreement with our prediction in the hypotheses. The minimum Euclid distance from the balance loss region is defined as the margin, and the margin from forward balance loss rapidly decreased after the stumbling perturbation. The statistical results reveal that the margin from the forward balance loss region after perturbation is significantly smaller than the margin in steady state walking. These results suggest that balance map analysis provides a new aspect of walking balance, expanded for the stumbling and recovery behavior of human walking.The code for the balance map analysis is available at https://github.com/TakahiroKagawa/GaitAnalysis_BM.