Tendon-driven Continuum Robots with Extensible Sections - A Model-based Evaluation of Path Following Motions

Amanov, E; Nguyen, T -D; Burgner-Kahrs, J
International Journal of Robotics Research, 40 (1), pp. 7-23, 2021
Tendon-driven Continuum Robots with Extensible Sections - A Model-based Evaluation of Path Following Motions

Abstract


Continuum robots are highly miniaturizable, exhibit non-linear shapes with several curves, and are flexible and compliant. In particular, concentric-tube and tendon-driven continuum robots can be designed on a small scale with diameters of below 10 mm. A small diameter-to-length ratio enables insertion of these robots through small entry points in order to reach hardly accessible regions by avoiding obstacles. This scenario can often be found in minimally invasive surgery and technical inspections. However, to reach the target region, a deployment along a narrow tortuous path is often required. Common tendon-driven continuum robots are intrinsically incapable of such deployment and concentric-tube continuum robots require special path conditions and intensive parameter optimization. Other proposed robot types, such as hyper-redundant and pneumatically actuated robots, exhibit less favorable diameter-to-length ratios and are thus not suitable for those tasks. Since the limiting factors are found in the design of continuum robots, we propose a novel tendon-driven continuum robot design, which features an additional degree of freedom in each robot section. The backbone is composed of straight, concentrically arranged tubes, each of which composes a section and is used to adapt its length. We present a three-section continuum robot prototype with a diameter of 7 mm, determine its follow-the-leader capabilities theoretically, and validate the results experimentally using model-based control. For our 165 mm long robot prototype, the repeatability is below 2.38 mm. The model accuracy reaches a median of 3.16% over 25 configurations with respect to robot length. The path-following error over five curvilinear paths results in median errors of 2.59% with respect to robot length.

External Links

BibTex   DOI