Design of a bio-inspired adaptive central pattern generator control for pneumatic muscle antagonistic joint

This paper presents a novel control strategy for an antagonistic joint actuated by pneumatic artificial muscles. Unlike traditional approaches, where Central Pattern Generators primarily serve as trajectory planners requiring secondary control, the proposed framework employs them as direct controllers generating pressure signals. This design mimics the biological motor system, creating a more integrated and efficient control process. In contrast to conventional control strategies, the proposed method removes the dependency on internal dynamic model, addressing singularities, computational instability in inverse dynamics, and the high cost of inertia matrix and Coriolis force calculations. By streamlining the control architecture and eliminating the need for separate trajectory tracking controllers, a leakage-type adaptive mechanism ensures robust parameter convergence, simplifies the overall design, enhances real-time responsiveness, and exhibits strong generalization capability in handling diverse trajectories and operational conditions. Simulations validate its robustness, ensuring smooth, precise control under disturbances and noise. Furthermore, this research drives the advancement of biologically inspired robotic systems, offering autonomous, adaptable, and efficient solutions for real-time control applications.