Stable computed-torque control of robot manipulators via fuzzy self-tuning

Area of Science:

• Robotics and Control Systems
• Applied Mathematics

Background:

• Computed-torque control is a standard motion control strategy for manipulators.
• It guarantees global asymptotic stability with fixed proportional and derivative (PD) gain matrices.
• Practical robotic systems face constraints like joint friction and limited actuator torque.

Purpose of the Study:

• To demonstrate that global asymptotic stability can be achieved with state-dependent gain matrices in computed-torque control.
• To enhance the computed-torque control scheme's applicability to real-world robotic systems with practical constraints.
• To introduce a fuzzy self-tuning algorithm for adaptive gain selection.

Main Methods:

• Theoretical analysis to prove global asymptotic stability for state-dependent gain matrices.
• Development of a fuzzy self-tuning algorithm to dynamically adjust proportional and derivative gains.
• Implementation and experimental validation on a two-degrees-of-freedom robot arm.

Main Results:

• Global asymptotic stability is proven for computed-torque control with state-dependent gain matrices.
• The fuzzy self-tuning algorithm effectively adapts gains based on tracking position error.
• Experimental results confirm the approach's usefulness in handling practical robotic constraints.

Conclusions:

• State-dependent gain matrices extend the stability guarantees of computed-torque control.
• The proposed fuzzy adaptive control strategy improves robustness against friction and torque limitations.
• The approach offers a practical enhancement for real-world robot manipulator control.