Design Engineering
Showcase 2020

Over the years, a number of researchers have explored methods to create a manipulator to mimic an elephant trunk. These manipulators are now recognised as continuum robots, due to their continuous curving nature. This makes them incredibly flexible and dextrous, which has prompted research into its application to the NOTES (Natural Orifice Transluminal Endoscopic Surgery) method. However, this research quickly demonstrated that achieving precise control using a continuum robot was difficult. This was due to the kinematic redundancies and the tangential vectors creating inaccuracies in mathematical models. To tackle this problem, I have developed an open source continuum robot, which simplifies the problems and ensures future researchers can, instead, easily create experiment-based models.

In order to validate the design and demonstrate the successful implementation of Endo as an open source project, a set of assessment criteria was made:

• Endo should use accessible manufacturing methods.
• Endo should be made of low-cost, easily attainable components and parts.
• Endo’s design should be adaptable to provide users freedom when tailoring designs.
• Endo should be a fully functional continuum robot.

Endo was then built in three parts: the spine, the control mechanism and the control system.

The Spine (left) was designed to use 4 tendons to control each segment, where each tendon would be able to pull the segment in one of the cardinal directions. Using 4 tendons had structural benefits of reducing tendon stress and would provide better perception of the different tasks. The spine was built using ABS rigid discs connected by TPU ligaments which provided restoring forces to under-actuate the robot. A series of iterations were also developed to optimise the design of these ligaments and these parameters can now be adjusted in the final design using SOLIDWORKS’ equations sheets.

The control mechanism (right) utilises low cost 55 g servo motors with 3D printed pulley attachments. The dimensions of these pulleys were dependant on which segment they needed to actuate, as segments further away from the robot needed to compensate for changes occurring in closer segments. This compensation was computed by the control system.

The control system was developed by creating mathematical relationships to calculate the approximate actuation ratios that would manoeuvre the spine as desired. This was implemented using Arduino and a custom library was built with different control functions, which provided increased amounts of control.

These parts were then assembled into Endo – the continuum robot system and its functionalities, and control functions were assessed by exploring the different configurations (left) and plotting the workspace (right). Finally, the designs and Arudino Library were uploaded to Github.