top of page

Robotics Design, Embedded 3D Printing, Path Optimization

Concentric Push-Pull Tube Robot for Embedded 3D Printing

A universal continuum robot platform with two concentric push-pull tubes for embedded 3D printing, developed for Skylar-Scott Laboratory at Stanford University.

Awarded David M. Kennedy Prize for Best Thesis in Stanford Engineering.​ Read thesis here.

Screenshot 2026-02-16 at 10.58.20 AM.png
IMG_6337_edited.jpg
Screenshot 2026-02-16 at 10.56.49 AM.png
Screenshot 2026-02-16 at 11.01.21 AM.png

Objective

To design, fabricate, and calibrate a precise robotic system for embedded 3D printing using concentric push-pull tubes. 

Outcome

The robot achieves sub-millimeter precision and has unique capability to print within nested, occluded geometries.

Project Background

image.png
Screenshot 2026-02-02 at 2.03.58 PM.png
image.jpeg

Image from Skylar-Scott Lab of bioprinted vasculature network (Andrew Brodhead/Stanford)

Organ Printing Need

Organ shortages continue to outpace demand. 3D bioprinting offers a path to patient-specific tissues that could reduce rejection and donor reliance.

Embedded 3D Printing

Embedded 3D printing extrudes ink into a gel bath that supports soft materials, enabling complex structures without requiring printing supports.

Image from Skylar-Scott Lab of bioprinted heart tissue (Andrew Brodhead/Stanford)

Current Limitations

Current bioprinting methods lack both precision and speed, posing challenges for print fidelity and cell survival that hinder large scale printing capabilities. 

Robotic Design

Design Strategies

  • Field research to gain inspiration from existing designs

  • Decision matrices to evaluate mechanism and component selections 

  • Breaking the design into subsystems... how can I control the motion of a single tube, and then how can I integrate this as a single unit in a larger system?

  • Consulting mentors, experienced designers, and field experts

Parallel Actuation Unit

Designing the system as a parallel actuation unit allows each platform to control the independent translation and rotation of a single tube, with each platform existing as its own subsystem for simpler design. 

Screenshot 2026-02-02 at 2.55.40 PM.png
Screenshot 2026-02-02 at 2.54.54 PM.png

Individual Platform Design 

The lead screw controls the translation of the platform, and the belt system drives rotation of the pulley. The tubes mount on the pulley using a shaft collar and tube mount to allow for the tubes to be easily swapped out of the robot.

Platform Integration 

Three platforms translate along linear rails to form the entire robot structure, with each platform controlling the independent translation and rotation of one tube. All motion is controlled by Nema 14 stepper motors connected to an Aerotech control motion system.

Screenshot 2026-02-02 at 10.23.08 AM.png

Push-Pull Tube Design

Working Principle

Concentric push-pull tubes feature a pair of inner and outer notched tubes which are fixed together at one end. As the other end of the inner tube is pushed or pulled relative to the outer tube, the CPPR bends in either direction.

Screenshot 2026-02-15 at 3.37.17 PM.png

Optimization Script 

To determine the optimal notch parameters for the push-pull tubes to achieve the desired curvature, I developed a MATLAB optimization script based off the algorithm developed by Oliver-Butler et al (shown in left image).

The full optimization script can be found on Github here, but from a high level: 

Script Inputs: 

  • Desired curve equation

  • Outer diameter of tubes

  • Wall thickness of tubes

  • Length of tubes

  • Number of notches

  • Tube flexural modulus & flexural strength

Script Output: Optimized notch geometry

  • Notch depths

  • Notch lengths

  • Distance between notches

  • Plot of resulting curve geometry

Tube Material & Fabrication

I began this project with the intention of fabricating the tubes using nitinol, a nickel-titanium alloy with superelastic behavior that allows the tubes to bend and return to their original shape. Due to limitations in nitinol laser cutting, I switched to 3D printing the tubes using Formlabs Tough 2000 resin, chosen for its stiffness, a crucial property due to the thin walls and small notched regions of the tubes. The print orientation of the tubes is shown right — after printing, the tubes were cured and the supports were snipped.

Screenshot 2026-02-16 at 12.40.31 PM.png
Screenshot 2026-02-15 at 3.39.21 PM.png

Tube Integration

The resin push-pull tubes are mounted to long steel tubes that traverse through the center of the robot and connect to their respective translating platforms. The print ink travels through a nitinol tube that runs through the center of the push-pull tubes and extends 20 mm past the end of the tubes to reduce dragging of the gel print bath (image A). The steel tubes attach to the translating platform by means of a clamping tube mount (images B and C).

Print Material Optimization

To develop the best material formulation to enable high-fidelity printing, I tuned the rheology of an epoxy print ink and a fumed-silica/mineral-oil support matrix to minimize path distortion and matrix crevices. To do so, I...

  • Performed rheometry testing (viscosity vs. shear rate and oscillatory modulus/yield behavior) across multiple matrix concentrations to identify viscosity-matched ink/matrix pairs.

  • Validated findings with experimental prints, iterating formulations to balance support bath stability and print accuracy—ultimately selecting a higher-viscosity ink with a more stable support matrix for best fidelity.

.

Screenshot 2026-02-15 at 3.40.42 PM.png
Screenshot 2026-02-15 at 3.41.50 PM.png

Calibration

Position Tracking 

I used Digital Image Correlation (DIC) software to track the position of the tube tip for each 1mm translation of the inner push-pull tube.

Based on the result of the position tracking, I developed equations for the r and z position of the tube tip as a function of the relative translation of the inner push-pull tube.

Screenshot 2026-02-16 at 1.37.15 PM.png
Screenshot 2026-02-16 at 12.04.17 PM.png
Screenshot 2026-02-16 at 12.04.28 PM.png

Calibration Results

Screenshot 2026-02-16 at 12.05.03 PM.png
Screenshot 2026-02-16 at 10.09.54 PM.png

I plugged the calibrated motion equations into a path planning script of a star shape, using a long exposure image and DIC to test the performance of the robot along corners and straight edges. The path stayed consistent across 10 trials, but shows hysteresis within a single star cycle.

I adjusted the motion equations to correct for this hysteresis by accounting for the previous path when calculating the robot position.

Printing Results

Print Setup

I mounted the robot to the Aerotech motion system, which controls the translation of the entire robot in the vertical direction, the translation of the inner tube, and the coupled rotation of both tubes to achieve the desired print coordinates. The extrusion of the print ink into the print bath is controlled by a pressurized air system, which plunges a syringe of print ink into the nitinol tube.

Screenshot 2026-02-16 at 10.55.50 AM.png

Print Shape

I chose to print a nested vase shape to highlight the unique ability of CPPRs to print in occluded geometries. A traditional linear nozzle system is not able to print both vases subsequently without colliding into the previous vase structure. 

Screenshot 2026-02-16 at 12.04.05 PM.png

Printing Process

A) MATLAB plot of 2D vase curvature, to be revolved into a 3D structure

B) Printing first few rings of outer vase

C) Completed outer vase print 

D&E) Inner vase printing

F) Final printed structure

Screenshot 2026-02-16 at 12.05.46 PM.png

Outer Vase Printing at 60x Speed

Inner Vase Printing at 60x Speed

bottom of page