Design, Optimisation, and Actuation of Tendon-Driven Hyper-Redundant Manipulators

Tendon-driven hyper-redundant manipulators allow safe inspection and maintenance in confined, hazardous environments by keeping actuators outside the work zone and moving through tight, cluttered spaces. However, their long tendon routings introduce friction, elasticity, and slack that often degrade performance in real-world scenarios. This thesis addresses these limitations through a complete design-to-application framework. A lightweight, 3D-printed 5-joint polymeric manipulator is developed with compact integration, embedded sensors, and an off-board actuation unit that includes active global tendon tensioning. The manipulator provides a platform for investigating real-world performance limits. A central challenge in the robot’s design was determining critical load cases for design optimisation. Traditional worst-case loading assumptions led to overdesign and inefficient material use. To overcome this, a data-driven framework based on Gaussian Process regression and Bayesian Optimisation is introduced. It identifies load configurations that induce worst-case stress on structural components while minimising the number of costly simulations. This method allows both design-time optimisation and safe operation by predicting and avoiding stress-critical joint configurations. Using this platform and optimisation framework, experiments are conducted on the robot to quantify tensioning efficiency at rest and in motion, while systematically excluding joint configurations associated with maximum stress. Furthermore, a simple model is obtained to link commanded tension to slack mitigation. Subsequently, experiments quantify hysteresis losses and overall mechanical efficiency of the tensioning system. Further testing characterises stiffness modulation and establishes a direct relationship between tendon tension and payload capacity. Finally, the optimisation framework is applied to design a longer underwater variant with 8 joints, adapted for inspection tasks in underwater environments. Buoyancy and payload constraints are addressed through geometry reconfiguration and structural mass reduction, allowed by the proposed design method. Overall, the thesis provides a generalisable approach to tendon-driven robot design, combining mechanical design, experimentation, data-efficient optimisation, and adaptation to real-world scenarios.