ARCSnake

Four years on an amphibious, screw-propelled snake robot developed with NASA JPL, spanning my first paper to the full-system publication at IROS 2026, which was co-led by Sara Wickenhiser and Elizabeth Peiros (I am third author). Screw propulsion is one of the few locomotion methods that works across sand, water, gravel, and mud without changing hardware, which is why it was proposed for exploring the subsurface ocean of Saturn’s moon Enceladus.

I also led electrical and mechanical integration for validating the full robot in underwater testing.

System design improvements

Across ARCSnake V2’s development I worked on the propulsion system, increasing screw output torque by 40%. The gains came from three places: raising the torque ratio, aligning the belt drive, and cutting parasitic friction, shortening the whole drive stack and relieving surfaces that were rubbing. I built a physical testbed to validate each change rather than trusting the CAD.

The screw block drive: the belt (green) transmits motor torque to the screw. Aligning this drive and shortening the stack around it is where the torque gains came from.

The screw testbed — Mobility Analysis of Screw-Based Locomotion and Propulsion in Various Media

To find out which screw parameters actually matter, I engineered a mobile testbed that isolates a single screw sub-unit and measures its performance across media: constrained axial measurement with a 6-DOF force-torque sensor, a swappable bottom module so different screw configurations drop straight in, built-in electronics bays and wiring paths, and SolidWorks FEA validation before fabrication. It is portable enough to run in real terrain rather than only in a lab.

Screws were characterized across gravel, grass, sand, wood chips, dirt, and concrete, in single and paired configurations.

Mechanical advantage against input torque in each medium, error bars across repeated runs. The useful finding is the shape, not the peak: screws are inefficient at low torque everywhere, and most media only cross the 25% efficiency line above roughly 1.25 Nm. That argues for gearing a screw robot to sit high on this curve rather than sizing it for the lightest expected load, and it is why grass and dirt, which never get there, are the cases where shell geometry matters more than raw torque.

That characterization became Mobility Analysis of Screw-Based Locomotion and Propulsion in Various Media (co-first author, ICRA 2023). The testbed has since been reused by the lab for further screw-locomotion studies beyond that paper.

NASU: ISRR 2024

NASU (Novel Actuating Screw Unit) is the first Archimedes-screw locomotion design with a dynamically reconfigurable angle of attack, letting one screw retune its pitch for the medium it is currently in rather than compromising across all of them. The mechanism is origami-inspired: a Kresling unit produces the coupled rotation and translation that changes the screw pitch (first author, ISRR 2024).

NASU mechanism overview.

The Voodoo Doll

A 10-DOF joint-matching teleoperation controller I built to command ARCSnake V2’s joint positions, lockable U-joints with magnetic encoders, so an operator poses the controller and the robot mirrors it.

The full 10-DOF controller laid out.