Design of a Hierarchical Control Architecture for Fully-Driven Multi-Fingered Dexterous Hand

Area of Science:

• Robotics
• Control Systems
• Biomimetics

Background:

• Dexterous robotic hands utilize high degrees of freedom (DOFs) and biomimetic designs for complex tasks.
• Pneumatic artificial muscles (PAMs) mimic human tendons and muscles, offering compliant and adaptable features in robotic hands.
• Nonlinear and hysteresis behaviors of PAMs pose significant challenges for real-time control accuracy and stability.

Purpose of the Study:

• To propose a hierarchical control architecture for optimizing the nonlinear control of PAMs in dexterous robotic hands.
• To address the challenges associated with PAM nonlinearity and hysteresis for improved robotic hand performance.
• To develop a robust control system for precise and stable operation of multi-fingered robotic hands.

Main Methods:

• Implementation of a hierarchical control architecture featuring a fuzzy PID strategy.
• Integration of an FPGA-based hardware system with a multi-channel DAC and multiplexed acquisition module for independent actuation and real-time monitoring of 20 PAMs and 20 joints.
• Software development incorporating a fuzzy PID algorithm for dynamic adjustment of PID parameters based on error and error rate.

Main Results:

• High precision achieved in controlling individual finger joint angles, with errors within ±1°.
• Excellent synchronization and real-time performance demonstrated in multi-finger cooperative grasping and biomimetic motion tasks.
• Validation of the fuzzy PID control strategy's efficacy in managing PAM nonlinearities.

Conclusions:

• The proposed fuzzy PID control system effectively manages the nonlinear behaviors of PAMs in dexterous robotic hands.
• The system meets precision and stability requirements for complex operational tasks, supporting the application of PAM-driven multi-fingered hands.
• The developed control architecture provides a robust solution for advanced robotic manipulation.