• Journal of the Korea Society of Computer and Information
• /
• v.20 no.3
• /
• pp.89-96
• /
• 2015

IT is widely used for biomechanics and AHRS sensor also be highlighted with small sized characteristics and price competitiveness in the field of motion measurement and analysis of sports. In this paper, we attach the AHRS to the figure skate shoes to measure the motion data like spin, forward/backward, jump, in/out edge and toe movement. In order to reduce the measurement error, we have adopted the sensors equipped with Madgwick complementary filtering and also use Euler angle to quaternion conversion to reduce the Gimbal-lock effect. We test and experiment the accuracy and execution time of the pattern recognition algorithms like PCA, ICA, LDA, SVM to show the recognition possibility of it on the basic motions of figure skating from the 9-axis trajectory information which is gathered from AHRS sensor. From the result, PCA, ICA have low accuracy, but LDA, SVM have good accuracy to use for recognition of basic motions of figure skating.

• Journal of Integrative Natural Science
• /
• v.17 no.3
• /
• pp.113-122
• /
• 2024

Advancements in sensor technology, particularly Inertial Measurement Units (IMU), are crucial in modern pose estimation. IMUs typically consist of accelerometers and gyroscopes (6-axis), with some models including magnetometers (9-axis). This study investigates the impact of sensor frequency on pose estimation accuracy using data from a 256Hz IMU sensor. The data sets analyzed include "spiralStairs," "stairsAndCorridor," and "straightLine," with frequencies varied to 128Hz, 64Hz, and 32Hz, and conditions categorized as stationary or dynamic. The results indicate that sensitivity remains high at lower frequencies under stationary conditions but declines in dynamic conditions. Performance comparison, based on Root Mean Square Error (RMSE) values, showed that lower frequencies lead to increased RMSE, thus diminishing model accuracy. Additionally, the Extended Kalman Filter (EKF) was tested as an alternative to Madgwick's algorithm but faced challenges due to insufficient sensor noise data.