Fast and Robust Actuation of Continuum Robots via Antagonistic Tendons

Tendon driven continuum robots (TDCRs) exhibit extraordinary dexterity and navigability in highly constrained workspaces, for example, in minimally invasive surgery or industrial inspection and repair tasks. The curvilinear nature of these robots, with their extreme diameter-to-length ratios, enables them to dexterously navigate around obstacles in their environment to reach their goals. However, the dynamic properties of these highly deformable mechanisms make them challenging to operate autonomously at high speeds, and the presence of external loads can significantly impact the accuracy of kinematic models. In this work, we present a compact, low-cost, modular tendon robot actuation platform with fast and stable open-loop dynamics. We explore how pretensioned antagonistic pairs can enable repeatable actuation at high bandwidths using our low-cost hardware setup. Secondly, we show how intrinsic sensing using only motor currents and displacement can be used to improve the shape estimate of the robot when manipulating unknown objects at its tip, without any additional sensors. In this work, we show how high-performance continuum robots can be achieved with relatively low-cost components through the mechanical design of the actuation structure and by leveraging existing information available from the actuators in the computational modeling of these systems.