Case Western researchers tame spiral robot arms with new closed-loop control

A team from Case Western Reserve University has introduced the first morphology-specific closed-loop control framework for logarithmic-spiral continuum arms. These arms, inspired by biological appendages like elephant trunks and octopus tentacles, offer unique capabilities for reaching, wrapping, and grasping. However, their highly underactuated and nonlinear nature has previously prevented precise control. The new approach leverages the unique logarithmic-spiral geometry to derive an analytical task-space Jacobian, then combines it with an online Jacobian error compensation system that uses a Broyden secant update and Kalman filter estimation to continuously correct modeling errors from deformation, contact, and geometric mismatch.

The framework was validated through extensive planar and spatial simulations, including trajectory tracking, attitude regulation, disturbance rejection, and simultaneous position-orientation control. Compared to a traditional piecewise-constant-curvature (PCC) baseline, the proposed method consistently reduced tracking errors, suppressed attitude drift, and maintained a bounded Jacobian estimation error. The controller was also demonstrated on morphology-enabled tasks such as obstacle-assisted reach-wrap-release, adaptive whole-arm grasping, and cooperative multi-arm object handling. The work establishes a physics-grounded foundation for future hardware implementations and learning-augmented soft robotic control.