An architecture for measuring joint angles using a long period fiber grating-based sensor.

Area of Science:

• Biomedical Engineering
• Robotics
• Sensor Technology

Background:

• Real-time signal filtering requires balancing computational resources and system performance.
• Fiber-optic sensors offer potential for precise motion measurement.
• Developing autonomous robotic systems necessitates efficient and accurate sensor data processing.

Purpose of the Study:

• To implement a low-lag, resource-efficient signal filtering technique for fiber-optic sensor data.
• To utilize a Long-Period Fiber Grating (LPFG) sensor for measuring finger bending.
• To classify finger positions using Gaussian Mixture Models (GMM) for robotic hand control.

Main Methods:

• Employed the Recursive Least Square (RLS) algorithm for real-time signal filtering.
• Utilized a Long-Period Fiber Grating (LPFG) sensor to capture finger bending movements.
• Applied Gaussian Mixture Model (GMM) for classifying finger positions.

Main Results:

• The RLS algorithm provided low lag response and reduced resource consumption for LPFG sensor data.
• Accurate measurement of finger motion angles was achieved.
• The GMM successfully classified finger positions along the motion range.

Conclusions:

• The proposed RLS filtering and GMM classification technique is effective for processing fiber-optic sensor data.
• The system is suitable for real-time implementation on platforms like FPGA and DSP.
• This approach facilitates the development of autonomous robotic hands with precise finger motion control.