MiniRHex Makes Wiggly-Legged Unstoppability Tiny and Affordable

For about $200, you can build a surprisingly capable six-whegged robot with googly eyes

4 min read
Photo: CMU

RHex (pronounced “rex”) is a unique hexapedal robot that uses hybrid wheel-legs (whegs) to get around. It’s surprisingly adaptable, able to adjust its gait to conquer a variety of obstacles and terrains, and it can even do some impressive parkour. RHex has been around for nearly two decades, which is practically forever in robot years, but because of how versatile it is you still see it doing cool new stuff from time to time.

Carnegie Mellon University’s Robomechanics Lab uses a fancy US $20,000 version of RHex called X-RHex Lite “to explore the connection between dynamic locomotion and perception,” but they’ve only got one robot since it’s wicked expensive, which limits the amount of research and outreach they can do. To fix this, they’ve designed a much smaller version of RHex called MiniRHex that you can build yourself for about $200. And it’s adorable.

Wow. This is how to make a good robot video, folks.

MiniRHex weighs in at under half a kilogram, but can support a payload of up to 3 kilograms. Six Dynamixel XL320s power the legs, driven by a ROBOTIS main board that talks to your computer via Bluetooth. Most of the structure of the robot is 3D printed, which keeps the cost quite low: If you have access to a 3D printer and a laser cutter, the entire robot will run you just over $200, or around $250 if you also need to buy the Bluetooth module and a charger for the battery. There’s a tiny amount of soldering plus some software setup that doesn’t look too difficult, and the instructions seem very easy to follow.

As you can see from the video, MiniRHex can, with a little bit of work, clamber over obstacles at least as high as it is, and it can scamper along at several body lengths per second. These aren’t optimized gaits either—while MiniRHex can currently take advantage of an alternating tripod gait as well as a pronking gait, there’s still plenty of room for optimization. Beyond just tweaking the gait in software, the size and springiness of the legs themselves can be adjusted as well, which is one of the reasons why RHex platforms are so interesting to work with. Here’s some preliminary gait testing with MiniRHex on a treadmill; watch until the end for a few outtakes.

For more details on MiniRHex, we spoke with Aaron Johnson, who runs the Robomechanics Lab at CMU, via email.

IEEE Spectrum: RHex has been around for a very long time. Why is it still a relevant and exciting robotics research platform?

Aaron Johnson: RHex is still relevant because it is one of the simplest walking robots and is easy to control. The RHex project started around 1999 and was inspired by the cockroach’s ability to get over rough terrain without careful footstep planning. Similarly, RHex’s compliant limbs and alternating-tripod gait make it very good at walking or running over obstacles without the need for slow, precise sensing, all while requiring only a single active degree-of-freedom per leg. The full size RHex is also perfect for climbing human-sized stairs, which is a limitation of many robots. Finally, RHex’s morphology make it great as a payload machine, since the weight of any sensors or other instruments is carried primarily in the material of the legs and not the motors. Cockroaches aren’t going away anytime soon, and I don’t think RHex will either.

Why make a smaller RHex?

The initial motivation was cost. We wanted to be able to come into a classroom of high school students with a fleet of robots so that they could have more hands-on time with the robot. MiniRHex is about 100 times lower cost than the full size platforms. But we have since found that there are advantages to being small. The scaling laws for material properties are such that MiniRHex can carry up to 6 times its own body weight, which for RHex would be about 120 lbs, even though the legs are 3D printed. There are also few legged robots that size that have independent control of each leg, and so it provides us with a small robot that we can still ask interesting questions about. One project we have right now is exploring different ways to use machine learning to develop walking and running behaviors on the robot. It is much easier to run these experiments on MiniRHex on a treadmill than it would be with a larger platform, allowing us to test these algorithms more easily. 

“One project we have right now is exploring different ways to use machine learning to develop walking and running behaviors on the robot. It is much easier to run these experiments on MiniRHex on a treadmill than it would be with a larger platform, allowing us to test these algorithms more easily”

Can MiniRHex do pretty much everything that RHex can do, just at a smaller scale?

For the most part yes. It still has six independent legs, so we can program in any gait patterns we would like. The top speed of the motors is a little slower than I would like, so it can just barely run but at a max of about 2 body lengths per second (compared to 5+ for RHex). However, we are working on a high performance version that should be able to really sprint. The size also means that while it can climb scaled-down staircases, almost all staircases are human size so the practical value of the behavior is not as great.

What you hope to be able to do with a swarm of MiniRHexes?

In addition to our outreach efforts with multiple robots, I think the best advantage of the RHex morphology, big or small, is the payload capacity. We hope to be able to deploy a swarm of MiniRHexes with a variety of sensors to be able to track chemical signals, measure environmental changes, or search for objects.

All of the detailed info on MiniRHex, including a parts list with links, assembly instructions, and operating instructions, can be found at the links below.

[ MiniRHex ] via [ Robomechanics Lab ]

The Conversation (0)

How the U.S. Army Is Turning Robots Into Team Players

Engineers battle the limits of deep learning for battlefield bots

11 min read
Robot with threads near a fallen branch

RoMan, the Army Research Laboratory's robotic manipulator, considers the best way to grasp and move a tree branch at the Adelphi Laboratory Center, in Maryland.

Evan Ackerman

“I should probably not be standing this close," I think to myself, as the robot slowly approaches a large tree branch on the floor in front of me. It's not the size of the branch that makes me nervous—it's that the robot is operating autonomously, and that while I know what it's supposed to do, I'm not entirely sure what it will do. If everything works the way the roboticists at the U.S. Army Research Laboratory (ARL) in Adelphi, Md., expect, the robot will identify the branch, grasp it, and drag it out of the way. These folks know what they're doing, but I've spent enough time around robots that I take a small step backwards anyway.

This article is part of our special report on AI, “The Great AI Reckoning.”

The robot, named RoMan, for Robotic Manipulator, is about the size of a large lawn mower, with a tracked base that helps it handle most kinds of terrain. At the front, it has a squat torso equipped with cameras and depth sensors, as well as a pair of arms that were harvested from a prototype disaster-response robot originally developed at NASA's Jet Propulsion Laboratory for a DARPA robotics competition. RoMan's job today is roadway clearing, a multistep task that ARL wants the robot to complete as autonomously as possible. Instead of instructing the robot to grasp specific objects in specific ways and move them to specific places, the operators tell RoMan to "go clear a path." It's then up to the robot to make all the decisions necessary to achieve that objective.

Keep Reading ↓ Show less