https://penntoday.upenn.edu/news/penn-engineering-navigating-landscapes-lunar-robots-moon-and-mars

Blog / Detailed Engineering Timeline of past --> Now

Started with V1 hardware, proving our docking strategy, which included a turret with 3DOF, and a passive docking mechanism with 2DOF 

• SP25 - I had joined group and started developing the docking mechanism alongside Raymond, iterating over many versions of the docking mechanism, to have full compliance, while being robust, and easily mountable on other robots, as well as simple in mechnical design and could be quickly prototyped. 

During the spring, we had accomplished our final docking mechanism (V4), which has proven it's robustness in field-test ranging in sand-courts, beaches, and White Sands NM. It's final demo is to be done at AMES NASA Center.

• S'25 - Developing next was the Rover. The rover has gone thorugh 4 major iterations. First iteration was a simple proof of concept (V1). Second iteration (V2) was quickly designed to test our main driving conditions/control/electronics and assumptions with 8020 Aluminum extrusion to then test in the NJ beach and sand courts. quickly we found ourselves with many issues including:

1. Bottoming out

2. Powertrain / Batteries being too weak

3. Lack of thermal control

4. "Bulldozing sand" and getting stuck

5. Inappropriate tires

Our next version (V2) was a major re-desing.Our main goal was to tackle problems shown 1-5.

To fix bottoming out, I had suggested to potentially allow us to have a modular ride height, in which would allow us double our previous height of V2, in order to tackle the bottoming out issue.

The lack of thermal control was a simple heat transfer problem. By understand the P vs. T curves of our motors, we could determine duty-cycles vs. time, and predict when the motors needed to cool down. This was crucial as it required us to know how to manage our thermal temps at White Sands NM. By adding white plastic cladding, we were also able to reflect the sun and push heat out of the robot.

For powertrain, since we needed modular height, a new modular powertrain which consisted of a chain sprocket design was a re-design, which was different from our V2 version which included a drive-by-wheel design, where the motor was directly attached to wheel. This required a complete custom made axle --> hub design in which I designed and hand machined alongside our GRASP machine shop in the lathe and mill. This required extensive design review and had to be replicable, as we needed at minimum 4 with 2 spares. We also re-designed our electronics from V2 to include a modular battery design in which we could ship to white sands. 

Shoveling was due to our flat face as can be shown by V2 - Beach pHusky I also developed a front-face dome, which was inspired by X-RHex field deployable bio-inspired robot, that allowed it to push through sand wihtout getting stuck. 

Tackling the tires was challenging. After very unsuccesful field trials, our V-notch "tractor tires" seemed to be counterintuitive in the sand. These worked best by digging into the sand and pushing off of it, whcih in sand, the shear force is the weakest. After further investigaing, I had started to reserach how dune buggies rode in the sand so effectively? I then realized the paddle-tire design. Afterwards, there was no paddle tire design that matched our very struct criteria, in which it had to be in certain dimension, diameter, etc. This led to desing a custom poly-eurethane paddle-tire, this required to test various known paddle-tire contours and adapat them into into a tire using a dove-tail mechanism, with a modular base, meaning we could swap between different paddle's if needed, in the case we weren't shearing enough sand etc. This allowed us to transverse thorugh sand, by leverageing the compressiveness of the sand, and shearing off of it, after long conversations with Granular media experts at Prof. Douglas Jerolmack, and reading about terra-mechanics, we spaced the paddles evenly in such a way that we estimated that the distance between them would give us enough surface area to compress the sand stiff enough to push off of it. This was a major and critical issue that turned out to be a major success for us at White Sands. 

Summary of white sands trip:

White sands, was our pre-Ames trip hardware readiness test. We wanted to show:

1. Multi-robot collaboration

2. Show failure modes of: Rover (pHusky) struggling in sandy terrain & our Quadruped (Spirit) with our payloads (Docking mechanism / Turret combos)

3. Show successful transversals of both the Rover (phusky), and Quadruped (Spirit), proving our hypothesis and research questions that quadrupeds navigating more "easily" thorough sandy terrain 

4. Show pHusky being rescued by Spirit in various slopes.

5. Show pHusky rescued by two spirits, a "TRUSS" link: Spirit (Penn) --> Rover (pHusky) --> Spirit (USC)

Learning from our field-expereince at white sands would be an understatement. We tackled several issues as listed through 1-5, late nights in the hotel room or quick fixes in the 100F weather. Through our experience, we were successfully able to show docking between two heteregenous robots in various slopes. We successfully proved our turret mechanism and our docking mechanism robustness by pushing a stuck rover in the sand by a our quadruped (spirit), to successfully "rescue" the rover, as it would have in the moon.

We carefully scouted our slopes, ranging from slopes that had been previsouly tested in lab, in which was extensively written in ou published paper titled: "TRUSSES - A multi-robot collaborative experimental platform" International Symposium on Experimental Robotics (ISER 2025) · Jul 7, 2025.

What we learned: 

• Our preperation was good, we did not have too many critical issues during the tests in the field

• Testing at the beach / volleyball courts was crucial

• Logging of all data went great

• System's integration

• Rover re-design (V2 --> V3) was critical: Ride heigh/Powertrain/paddle tires

Future Work

• Software improvement. Need to test and implement autonomous driving, and inteagrate motion capture + GPS + Vision autonomy.

• Hardware improvements to Rover (as described before)

• Need to implement a custom pushing gait for quadruped > troit gate

• Implement turret control and autonomous docking

V4 - AMES NASA pHusky - Fall 2025

• Now we are at our V4, preparing for our Ames NASA demo in California. This is proven to be most difficult. From our V3 - "White sands pHusky", there were many lessons learned which included:

  1. Drivetrain not being stiff enough which caused unwanted camber in tires

  2. Removing robots from test-site proved difficult

  3. Issues with unwanted drivetrain wobble

  4. Sandproofing

  5. Modularizing battery swapping

  6. PCB electronics

Now we are tackling these issues, which are causing a major re-design of pHusky V4. First, we are tackling the motor issue, by supporting the V2 of the drivetrain axle --> hub, to be supported by two bearings, and mounting the drivetrain to the inside of the robot, so it could be easily swappable. 

Removing robots is being solved by adding handles that will act both as mounts to carry the robot, and insert poles to lift the robot. 

Sand-proofing is currently being researched, and we are thinking of implementing seals all around the robot between all seems, to improve sand-proofing, as well as better sand-proofing our quadruped, docking mechanism, and turret. 

Our electronics worked great, but all was hand soldered in proto-boards, our next goal is to design the circuits and print them (PCB) for robustness. 

Current Hardware Versions: (10/2025)

Left to Right: 2xRhex , First iterations of docking mechanism, PsuedoHUSKY ( Ground Rover) 

Multigait - Soft Robotics

Introduction:

•  Utilizes silicone-based materials for creating molds, actuators, and air channels.

•  Enables the creation of robots that can safely interact with humans.

•  Designed for tasks that traditional, hard robots cannot perform.

  •   Manufacturing: Soft robots can work alongside human operators with minimal risk.

  •   Search & Rescue: Capable of navigating tough conditions to handle fragile objects in challenging environments.

Summary:

•  Inspired by biological and animal mechanisms.

•  Initial inspiration came from a pneumatic robot with an elephant trunk made of silicone.

•  This led to the development of a robot inspired by a four-legged creature with multiple gaits.

•  Design and testing based on mathematical models for circular motion and analysis.

•  Features four air-actuated legs and a central chamber.

• Extensive fluid and mechanical analysis to optimize air channels and gaps for precise movements.

• Aims to explore various air channel designs, wall thicknesses, and silicone materials to enhance the capabilities of silicone-based robotics.

Current Work / Progress:

•  Focusing on finalizing wall thicknesses, air channel dimensions, and sizing for specific silicone types.

•  Utilizing 00-30 silicone, differing from the 00-50 silicone commonly used in other soft robotics.

•  Incorporating an Optitrack system and DC motors to develop a state-based control system and PID controller.

  • This system will enable the robot to perform tasks such as turning, moving in specific directions, and using all four legs for grappling.

•  Planning to publish findings with my Principal Investigator before my graduation from the University of Rhode Island in May 2024.

ROS (Robot Operating System) Simulations of ground robot with Lidar sensors mapping environment utilzing low level controllers with active obstacle avoidance.