✕ Close
Close

Contact

Feel free to reach out about research, projects, or anything you found intriguing.

Click to copy Copied.

Open in mail

LinkedIn ↗ · CV

Skip to content
spiral body (TPU-70A)M1M2relax_ratio = 0.2passive slack = 0.2 × active pull123staged bringup
  1. 1Single-motor tracking
  2. 2Stall characterization
  3. 3Coordinated control
Cable-actuated soft body with control tuned experimentally before coordinated motion.

SpiRobControls (Drexel Zheng Lab)

Staged bringup produced visible directional actuation of a two-cable SpiRob, with relax_ratio = 0.2 and documented motor, cable, and print limits for the next lab iteration.

View the code github.com

Research intern

2025

Arduino, SolidWorks, TPU-70A 3D Printing, AccelStepper, MuJoCo, Isaac Lab, Python

01The problem

The SpiRob is a logarithmic-spiral soft robot actuated by fishing-wire cables, but the two-cable version did not come with an established control stack. The low-cost stepper motors had limited torque, the fishing wire stretched under tension, and the TPU-70A body could buckle if print settings were wrong. Before coordinated control was even possible, the hardware had to be characterized piece by piece.

relax_ratio = 0.2 was tuned experimentally to avoid over-tension

Bringup sequence moved from single motor tests to coordinated control

02What I did

I followed a staged bringup process instead of jumping straight to coordinated motion: single-motor step tracking first, then stall testing to find safe limits, then experiments to tune the passive-motor slack ratio, and finally the coordinated controller. I also printed a mini SpiRob before committing to the full-size TPU-70A print, and built MuJoCo and Isaac Lab environments in parallel so later control work would not depend entirely on physical hardware.

IR remote and Serial monitor commands both work in the same sketch

MuJoCo and Isaac Lab environments were built alongside the hardware

03How it works

The production sketch uses AccelStepper and MultiStepper to drive up to three 28BYJ-48 motors at once. The key control value was relax_ratio = 0.2: when the active motor pulls N steps, the passive motor releases 0.2×N steps of slack so the system does not over-tension and stall. That value came out of repeated experiments, not guesswork. The controller accepts both IR remote input and Serial commands, and the fabrication notes document the TPU-70A settings that avoided buckling. On the simulation side, I built both a MuJoCo XML model and a USD Physics cable model for Isaac Sim.

Isaac Sim cable model uses segmented spherical joints

TPU-70A print settings and low-infill failure modes are documented

04What happened

The hardware achieved visible directional actuation of the two-cable SpiRob, and the documented bringup process gives the next lab iteration a real starting point instead of a blank slate. Just as important, the repo records the system's current limits, motor torque, cable stretch, and missing restoring force, so future design decisions can start from actual evidence.

A mini prototype came before the 12-hour full-size print

Build real things. Make them worth using.

About Contact GitHub ↗ Scholar ↗ CV