Model-Free Control of Movement in a Tendon-Driven Limb via a Modified Genetic Algorithm

Area of Science:

• Robotics and Biomechanics
• Control Systems Engineering
• Computational Intelligence

Background:

• Tendon-driven systems offer advantages over torque-driven systems but face challenges due to their over-determined nature and posture-dependent actuation.
• Accurate modeling of tendon-driven systems, especially biological ones, is often difficult, posing significant control constraints.

Purpose of the Study:

• To develop a control strategy for tendon-driven systems that overcomes model uncertainty and complex dynamics.
• To enable autonomous trajectory following in systems with variable mechanical structures, inspired by biological mechanisms.

Main Methods:

• A modified Genetic Algorithm was employed to determine optimal tendon excursion values.
• The algorithm was designed to provide control inputs without prior knowledge of the system's exact model parameters or structure.

Main Results:

• The proposed Genetic Algorithm successfully generated tendon excursion values that allowed the limb to accurately follow a desired trajectory.
• Effective trajectory tracking was achieved without exposing the control algorithm to the underlying system model.

Conclusions:

• The modified Genetic Algorithm provides a robust solution for controlling tendon-driven systems with unknown dynamics.
• This method holds potential for enabling autonomous movement in biologically inspired robots with adaptable mechanical configurations.