Labeling and Detection of Underwater Ankle Motions in Breaststroke with Wearable Sensors

Flexible ankle motions directly affect the kick effect of breaststroke. Accurate detection of critical moments of ankles [Feet Kick Start (FKS) and Feet Kick End (FKE)] in breaststroke is significant for sports and underwater assistance. In this article, we propose a wearable sensing system that can be used to label and detect critical moments of ankles in breaststroke based on the idea of supervised pattern recognition. Using the optical system as a reference, the feasibility of the flex sensor as a label was verified by the land experiment. Combined with the label data from the flex sensor AD value, the model was trained using the inertial measurement unit (IMU) data on the foot and shank at normal swimming speed. Based on the classification model, critical moment detection experiments were conducted on three subjects with different swimming speeds (slow, normal, and fast). The time differences of FKS detection were-6.0, 0.7, and-7.5 ms, which accounted for-0.23%, 0.12%, and-0.39% of the corresponding period, respectively. The time differences of FKE detection were-20.7, 3.50, and-23.5 ms, which accounted for-0.90%, 0.29%, and-1.21% of the corresponding period, respectively. To compare with the recognition algorithm, we have studied the threshold detection method for four variables. The results show that the wearable sensing system and recognition algorithm proposed in this article can be used to label and detect critical moments of ankles in breaststroke effectively, and detection results have high accuracy and small time difference, as well as certain speed adaptability.