In this study, data from WaveX and two TrackX IMU (Inertial Measurement Unit) sensors produced by Cometa srl, which can generate consistent and reliable data underwater, were used to analyze a swimmer’s stroke cycle. The IMU sensors were placed on the wrists of the swimmer to capture tri-axial acceleration data. This data was key in segmenting individual strokes and providing a detailed analysis of each swimmer’s stroke cycle phase.
- Processing Accelerometer Data
- Purpose: The accelerometer data was the first to be processed as it was needed to split the signals from each lap into individual strokes.
- Filtering: The raw accelerometer data was filtered using a low-pass Butterworth filter of 4th order at 5Hz to eliminate high-frequency noise and ensure smooth data for further analysis.
- Stroke Segmentation
- Objective: After filtering, the next task was to segment the data into individual strokes. This was achieved by applying an existing algorithm capable of identifying acceleration patterns related to different stroke phases.
- Algorithm Validation: The algorithm used for stroke segmentation was based on research by Yuji Ohgi (Keio University). It identified each stroke based on the acceleration peak observed on the Y-axis during hand impact with the water.
Impact Acceleration Peak: The global minimum of Y-axis acceleration indicates the moment a new stroke begins, marking the hand’s impact with the water.
- Transition Detection Between Phases
- Transition Phases: The algorithm was also designed to detect the transition between the underwater and aerial phases of the stroke.
- X-Axis Analysis: The X-axis acceleration was specifically analyzed to identify the transition point between these two phases. According to Ohgi’s observations, this transition is marked by the second positive peak in the X-axis acceleration.
- Time Normalization
- Normalization Process: Time normalization was applied to facilitate accurate comparison between different strokes. This step ensured that the stroke data from various laps could be aligned in terms of time.
- Standardization: The strokes were normalized between 0% and 100%, with the average stroke length of each swimmer set as 100%. This allowed for standardized data comparison across different swimmers and laps.
- Outcome
The processed data provided insights into the swimmer’s stroke cycle, allowing for detailed analysis of both the underwater and aerial phases. IMU sensor data, stroke segmentation, and phase detection algorithms made it possible to assess performance and technique with precision.
This analysis technique offers a robust way to study and improve swimming techniques by clearly breaking down each phase of the stroke cycle.
Do you want to know more about our EMG systems?
