Advancing Tendon-Driven Robotic Systems: From Climbing Robots to String Actuators

Abstract

Tendon-driven mechanisms provide a range of benefits for robotic systems, particularly by allowing actuators to be mounted at the base of a manipulator and reducing its inertia. This thesis explores two projects that exploit and advance tendon-driven mechanisms: a wheeled-grasping hybrid climbing robot with modular tendon-driven grasping arms and a hybrid twisted-winching string actuator. Called CLIMR (Cabled Limb Interlocking Modular Robot), the novel climbing robot adapts to columns of varying diameters by adding or removing modular arm links. CLIMR also features capabilities like self-locking (the ability of the robot to stay on the column without power), autonomous grasping, and rotation around the column axis. Mathematical models describe conditions for self-locking, vertical wheeled climbing, and complete grasping of a column. Simulations and experimental results validate the proposed models. The insights from CLIMR are then extended into general design strategies for future developments of similar hybrid climbing robots, focusing on methods to inform design decisions and assess metrics such as adaptability. Ultimately, this work provides a comprehensive framework for designing hybrid climbing robots, highlighting the potential of autonomous solutions for environments where climbing tall structures is critical. Stemming from this climbing robot work is a novel actuator system combining a twisted string actuator (TSA) with a winch mechanism. Relative to traditional hydraulic and pneumatic systems, TSAs are compact but face limitations in stroke length and velocity. This TSA-winch system overcomes these constraints without risking overtwisting by providing both high displacement winching and high force twisting modes. The design features a rotating turret that houses a winch and a worm gear transmission driven by a through-hole drive shaft. Models are developed for the combined displacement and velocity control of this system. Experiments validate the open loop model as well as the closed loop model, which uses a conductive string feedback controller with a gain scheduling and control effort allocation scheme. For specific cases that require large displacement winching followed by high force twisting over several repeatable cycles, an alternate design sacrifices complete string state control and replaces a motor with passive automatic clutches to achieve a seamless transition between modes triggered by the string load. The models of the clutch torque thresholds for this version of the actuator are verified by experiments. Overall, this research contributes to the development of more versatile and efficient actuation systems for tendon-driven robotic applications.