A biomechanical model for encoding joint dynamics: applications to transfemoral prosthesis control

Area of Science:

• Biomechanics
• Robotics
• Biomedical Engineering

Background:

• Micro-controlled prosthetics require advanced control systems for natural movement.
• Predicting joint dynamics is crucial for seamless prosthetic limb adaptation.

Purpose of the Study:

• To present and test a framework for encoding joint dynamics into energy states.
• To predict future knee joint energy states (torque and velocity) without prior activity knowledge.
• To enhance micro-controlled prosthetics using embedded sensory data and mechanical principles.

Main Methods:

• Utilized kinematic and kinetic knee joint sensor data.
• Developed a framework to encode joint dynamics into energy states.
• Applied the framework to human subjects during walking (preferred and fast speeds).

Main Results:

• Joint energy states were consistently sequenced (75% consensus) based on mechanical energy transference.
• Subsequences reflected knee stability and energy dissipation requirements during gait.
• Simulations predicted future energy states with >80% accuracy 2% cycle in advance and >60% accuracy 4% cycle in advance.

Conclusions:

• The encoding algorithm successfully predicts future joint energy states.
• This approach holds potential for improving prosthetic limb control and responsiveness.
• Further research is warranted to explore its application in identifying diverse human activities and transitions for TFP control.