Design and implementation of a low power mobile CPU based embedded system for artificial leg control

Area of Science:

• Biomedical Engineering
• Robotics
• Machine Learning

Background:

• Artificial leg control demands high accuracy, real-time processing, and low power consumption.
• Existing neural-machine-interfaces (NMIs) face computational challenges for mobile applications.
• Developing efficient NMIs is crucial for advanced prosthetic limb functionality.

Purpose of the Study:

• To design and implement a novel neural-machine-interface (NMI) for artificial leg control.
• To address the computational challenges of high accuracy, real-time processing, and low power consumption in mobile NMIs.
• To demonstrate the superior performance of the developed NMI system.

Main Methods:

• Utilized architectural features of a mobile embedded CPU for the NMI's computation engine.
• Implemented a decision-making algorithm based on neuromuscular phase-dependent support vector machines (SVM).
• Conducted real-time experiments with an able-bodied subject using a 20 ms window increment.

Main Results:

• Achieved exceptional accuracy (99.94%) in real-time experiments.
• Demonstrated efficient algorithm execution with less than 11% processor load.
• Confirmed the NMI's capability for high-speed, accurate control of artificial legs.

Conclusions:

• The designed NMI effectively meets the stringent requirements for artificial leg control.
• The mobile embedded CPU and SVM-based algorithm provide a highly accurate and efficient solution.
• This NMI represents a significant advancement in prosthetic limb technology.