Estimation of ground reaction force direction in inline speed skating motion

This study proposes and verifies a method to estimate the ground reaction force (GRF) direction from supporting-leg posture during speed skating. We hypothesized that GRF direction can be accurately estimated from supporting-leg posture. Two estimation methods were developed under the assumption that the reaction force acts perpendicular to the shoe sole. One method used the coordinates of the front and rear blade ends and the ankle joint center (Est1). In contrast, the other used the coordinates of the front and rear blade ends and the knee joint center (Est2). The magnitude of each GRF component was estimated by multiplying the resultant force measured by the force plates (FP) by the corresponding unit vector components calculated in Est1 and Est2. GRFs measured using FP were adopted as true values to verify the proposed methods. Sixteen experienced university speed skaters performed straight and curved inline speed skating trials, during which three-dimensional kinematic and FP data were recorded. The root mean square error (RMSE) between FP measurements and estimates from Est1 and Est2 was calculated for each force component. The largest observed error was approximately 10% of body weight. Across all skating motions and force components, Est2 consistently exhibited smaller RMSE values than Est1. Based on these results, Est2 was selected for further verification. For straight skating and the right leg during curve skating, bias values for all components were less than 8% of body weight. For the left leg during curve skating, larger biases were observed before and after the right blade contact; however, applying regression-based corrections substantially reduced these biases. Intraclass correlation coefficient analysis indicated agreement between true and estimated values for the X' and Z' components across all skating motions. For the Y' component, several time points showed no agreement with the true values. Nevertheless, because the Y' component magnitude was very small, its influence on calculations such as joint torque or joint power of the supporting leg was minimal. Overall, Est2 is suitable for these calculations.