TRUSSES (NASA LuSTR): End-to-End Systems Integration

I lead end-to-end systems integration across the TRUSSES multi-robot platform—spanning mechanical design, low-level control, autonomy integration, and field validation. The work centers on building field-deployable heterogeneous robot teams that can dock, assist one another, and traverse granular terrain under real environmental constraints (heat, sand ingress, slopes, and repeated deployment cycles).

What I Owned

Across TRUSSES, my primary ownership has been:

• Mechanical design and manufacturing iteration (V1 → V4)

• Low-level actuation + ROS 2 integration for mobility and docking

• Autonomy integration hooks (vision / GPS / mocap)

• Field test planning, execution, telemetry, and logging workflows

• Hardware hardening for the NASA Ames demo

Docking System (V1 → V4)

We developed a probe–drogue-inspired docking system consisting of a 3-DOF active turret (probe) and a 2-DOF passive drogue. I co-led the docking mechanism development with Raymond Yang, taking the system from concept to a compliant, robust, modular mechanism that has been validated in realistic field settings.

My Contributions

• Owned the mechanical + manufacturing iteration across versions

• Developed push-lock latching behavior and the compliant spring/gear system

• Drove design decisions to balance:

  • high compliance (alignment tolerance)

  • robustness (impact and repeated docking)

  • modularity (easy mounting to different robots)

  • rapid prototyping and repairability in the field

Validation

The final docking mechanism (V4) has been validated across:

• sand courts

• beaches

• White Sands, New Mexico

We are currently hardening the full system for the NASA Ames demonstration in synthetic regolith (moon sand).

Rover Architecture: pHusky (V1 → V4)

After early field trials, we redesigned the rover from a proof-of-concept chassis into a field-ready platform. The rover has gone through four major iterations, each driven by lessons learned during real deployments in sand and heat.

The Problems We Hit in Early Field Tests (V2 Era)

• Bottoming out in sand / terrain discontinuities

• Underpowered drivetrain and limited battery modularity

• Thermal limits in hot environments

• “Bulldozing” sand due to flat front geometry

• Tires optimized for digging (bad in loose sand)

The Solutions I Built (Key V2 → V3 Redesigns)

Modular Ride Height (≈ 2× Clearance)

I proposed and implemented a modular ride-height system to address bottoming out, enabling significantly more clearance without sacrificing serviceability.

Powertrain Redesign: Direct Drive → Chain–Sprocket

We replaced direct drive with a chain–sprocket drivetrain and a custom axle → hub assembly. I designed and hand-machined the axle/hub components (lathe and mill in the GRASP shop), with replication in mind (4 active + 2 spares).

Thermal Modeling + Mitigation for >100 °F Operation

I modeled motor power–temperature (P–T) behavior to define duty-cycle envelopes and predictable cooldown windows. We also added white cladding to reduce solar heating during high-temperature field operation.

Domed Front Geometry (Inspired by X-RHex)

To reduce plowing and “bulldozing,” I designed a domed front end inspired by field-deployable bio-inspired platforms (e.g., X-RHex). This significantly reduced the tendency to push sand and stall.

Batteries + Electronics Reliability

I introduced modular, field-swappable battery packs and pushed the electronics stack from protoboards toward more reliable PCB-based layouts.

Sand Mobility: Custom Paddle Tires

Our initial V-notch “tractor” tires performed poorly in sand because they relied on digging and shear—exactly where loose sand is weakest. I researched dune-buggy paddle tire designs and terramechanics, consulted Prof. Douglas Jerolmack (granular media), and designed custom polyurethane paddle tires.

What Made Them Different

• Modular dovetail paddle inserts with tunable spacing and contour

• Designed to balance compaction vs. shear, so the rover could build a stiff enough “push-off” layer rather than just excavating

This became a major enabler for reliable traversal at White Sands.

Autonomy + Control (ROS 2)

On the software side, I built low-level actuation and ROS 2 integration for docking and mobility, and led integration with higher-level autonomy components. I also implemented the system hooks needed to incorporate:

• vision-based tracking

• GPS integration

• motion capture pipelines

Field Validation: White Sands, NM

White Sands served as a hardware-readiness and systems-integration test before NASA Ames. The goals were to demonstrate:

• Multi-robot collaboration under real terrain constraints

• Rover failure modes in sand (and recovery strategies)

• Quadruped-assisted rescue behaviors (Spirit pushing/pulling pHusky)

• A TRUSS link demonstration: Spirit → pHusky → Spirit

• Robust docking between heterogeneous robots on slopes

We also established reliable telemetry and logging workflows, and validated that early “soft” testing in sand courts and beaches was essential preparation.

Current Work: V4 Hardening for NASA Ames (Fall 2025)

The V4 rover is focused on hardening and repeatability:

• Dual-bearing drivetrain supports to eliminate axle wobble and tire camber drift

• Internalized drivetrain mounts for stiffness and serviceability

• Sand-seal strategy across seams to reduce ingress

• Carry/lift handles for extraction and safe transport

• PCB power distribution for reliability and faster field repair

• Integration of autonomous driving + turret control

• Developing a custom pushing gait for Spirit to improve rover rescue performance in sand

What’s Next

• Expand autonomous driving validation in outdoor settings

• Full integration of vision/GPS/mocap pipelines into field autonomy

• Continued sandproofing + drivetrain stiffening improvements

• Robust autonomous docking demonstrations at Ames