Adaptive Higher-Order Sliding Mode Control of a Tendon-Driven Soft Joint

This thesis presents the design, modeling, and control of a tendon-driven soft robotic joint integrated into the arm of a social robot named AKI. The aim is to enable expressive and safe hand gestures for child-robot interaction. A Higher-Order Sliding Mode Controller (HOSMC) with and without self-tuning adaptive gains is implemented and evaluated based on accuracy and smoothness in motion. Additionally, the effect of replacing a rigid end-effector with a soft joint on the kinematic manipulability of the robotic arm is analyzed. Experiments were conducted by having the controllers follow sinusoidal trajectories with different amplitude, frequency and loads. The RMS position error and the SPARC value were measured for all tests. The experimental results indicate no statistically significant difference between the adaptive and standard HOSMC in terms of performance. However, the addition of the soft joint increases the manipulability ellipsoid volume by approximately 7 %.