Motor noise removal for determining gait events over treadmill walking using wavelet filter

Area of Science:

• Biomechanics
• Signal Processing
• Instrumentation

Background:

• Low-pass filtering is standard for force plate data but can be inaccurate with custom instrumented treadmills.
• Motor noise from treadmills contaminates ground reaction force (GRF) data.
• Accurate GRF data is crucial for biomechanical analysis.

Purpose of the Study:

• To compare the effectiveness of a wavelet filter against a conventional low-pass filter for processing force plate data from an instrumented treadmill.
• To assess the ability of a new convolution wavelet (CNW) filter to remove motor noise.
• To evaluate the performance of the CNW filter, particularly under conditions of high motor speed and low signal-to-noise ratios.

Main Methods:

• Collected ground reaction force data using force plates on a custom-made instrumented treadmill during operation.
• Applied a novel convolution wavelet (CNW) filter to the collected force plate data.
• Compared the results of the CNW filter with data processed using a conventional low-pass filter and band-pass filtering.

Main Results:

• The CNW filter successfully eliminated motor noise present in the force plate data.
• The CNW filter yielded more accurate force plate data compared to the low-pass filter, especially at high treadmill speeds.
• The CNW filter demonstrated superior performance to band-pass filtering, particularly in low signal-to-noise environments, and had a lower computational load.

Conclusions:

• The convolution wavelet (CNW) filter is a more accurate and efficient method for processing force plate data from instrumented treadmills than conventional low-pass filtering.
• The CNW filter effectively removes motor noise, improving data quality for biomechanical analysis.
• This advanced filtering technique is particularly beneficial for high-speed treadmill applications and noisy data environments.