Adaptive Tip Trajectory Generation for Zig-Zag Soft Actuators using Modular Strain-limiting Layer Design

Soft pneumatic actuators (SPAs) made from compliant materials are increasingly being used as alternatives to rigid robotic actuators, particularly in applications requiring greater flexibility, adaptability, and compliance. A typical SPA incorporates a single strain-limiting layer (SLL), which helps maintain the actuator’s structural integrity and regulates its deformation to control motion. These SLLs are constructed from materials with specific elastic moduli, causing the SPA’s tip to follow a predetermined trajectory. However, in dynamic environments where varied tasks such as versatile in-hand manipulation are required, a single trajectory is insufficient. To address this, we present a novel modular press-fittable SLL design for Zig-zag SPA configurations, enabling multiple trajectory paths without the need for repeated actuator re-fabrication. By varying the SLL configuration, different motion trajectories can be achieved, without altering the primary structure of the actuator. Additionally, we tackle the inverse kinematic problem of determining the appropriate SLL configuration for a desired trajectory, using support vector machines (SVM) to predict the optimal configurations. Our experiment demonstrates the ability to generate over 100 unique trajectories using only Dragon Skin 10 & EcoFlex 30, with the SVM model achieving an average accuracy of 95% in predicting the necessary configurations for specific trajectories.