Stringman: an $800 open source robot that hangs in your living room

A robotic arm that can reach across a whole room usually costs as much as a car. Stringman does almost the same thing for about $800, and the design is open. It's an open source robot with no arm at all. It has cables. It holds a gripper in the middle of the room and controls its position by pulling and releasing wires tied to the corners.

It looks like a hack. But it's one of the oldest and most elegant ideas in robotics, now packaged as a kit that fits on your credit card. The project lives on the Neufangled website.

What Stringman actually is

Think of the camera that flies over the field during football broadcasts. It hangs from four cables, each one attached to a motor high up in the stadium. When one motor reels in its cable and another lets out, the camera moves. There's a name for this: a cable-driven parallel robot.

Stringman takes that idea and brings it indoors. You mount the anchors up high, near the corners of the room. Cables run from them to a central module with a gripper. The motors control the length of each wire, and the gripper goes wherever the geometry says.

The result is a robot with a workspace the size of the room. A traditional benchtop arm reaches maybe half a meter. Here, the "arm" goes from one wall to the other.

Sometimes the cheap way to reach far is to stop building an arm and start stretching a rope.

Why a cable robot is so much cheaper

The cost of a robotic arm comes almost entirely from its structure. Each joint has to hold the weight of everything after it. That takes strong motors, precise gearing, rigid metal. And the longer the arm, the worse it gets: torque at the shoulder grows with distance.

With cables, the problem moves somewhere else. The motors stay put at the anchors. The moving part is light: the gripper and almost nothing else. The weight is spread across the wires, and the wall holds the rest.

In practice, you trade:

• Rigidity for reach. A cable can't push, only pull. The gripper sways a bit.
• Millimeter precision for a huge area. Don't expect to solder boards with it.
• Expensive structure for annoying installation. Someone has to mount anchors high on the wall and calibrate.

It's an honest trade. To pick a toy off the floor, carry a mug across the room or gather laundry, you don't need surgical precision. You need reach and a price that doesn't scare you.

How you control a robot hanging from cables

This is the part that matters if you write code. To put the gripper at a point (x, y, z), you need to know how much cable each motor should let out. That's inverse kinematics, and the cable version is surprisingly simple: the length of each wire is the distance between the anchor and the attachment point on the gripper.

type Vec3 = { x: number; y: number; z: number }

const dist = (a: Vec3, b: Vec3) =>
  Math.hypot(a.x - b.x, a.y - b.y, a.z - b.z)

// cable length each anchor needs for the gripper to reach `target`
function lengths(anchors: Vec3[], target: Vec3): number[] {
  return anchors.map((a) => dist(a, target))
}

That's it, that's the basic math. On a six-joint arm, inverse kinematics already forces you to solve nonlinear systems, deal with singularities and pick between several solutions. Here it's Pythagoras in 3D.

The devil is in the details, of course. Cables stretch. The spool changes diameter as it winds, so "one motor turn" isn't always worth the same amount. If a cable goes slack, the gripper loses stability. And the anchor positions need to be measured properly, or the whole calculation starts out crooked. That's where calibration comes in, and the project uses cameras to help the system know where the gripper really is.

The fun part: with a camera and software control, the robot becomes a platform for teleoperation and for training imitation learning policies. You guide the robot a few times, record the data and try to teach a model to repeat the task. It's the same path a lot of people are taking with cheap benchtop arms, just with a whole-room workspace.

Open source hardware: what changes for devs

We already get free software. You clone, run, open a PR. Open hardware is a different beast. Open code helps, but what really unlocks it is having the part files to print, the bill of materials and the step-by-step assembly guide. Without that, "open source" is just a pretty YouTube video.

When a robotics project opens everything up, three things happen:

1. The price becomes a ceiling, not a floor. If you already have a 3D printer and a spare board, the cost drops.
2. Bugs become everyone's. Someone finds out the spool slips, posts the fix, and the next person never runs into it.
3. The robot becomes a platform. People who would never design the mechanics can still work on the control, the vision and the AI side.

That third point is what interests me most. Most of us, web and mobile devs, don't know how to calculate motor torque. But we do know how to write a server that takes commands, a UI that shows a live camera feed, an app that sends the gripper to a point you click on screen. A cheap, open robot is the way in for people coming from software.

Is it worth building one?

It depends on what you want.

If the idea is to have a butler that tidies the house on its own, hold your excitement. A cable robot has clear physical limits. Cables crossing the room run into furniture, ceiling fans and, for sure, the cat. Payload is limited. And autonomy depends on software and data that are still far from "go wash the dishes".

If the idea is to learn real robotics without spending the price of a motorcycle, then it makes a lot of sense. You get to work with:

• kinematics you can understand in an afternoon;
• calibration and the clash between the ideal model and the real world;
• computer vision to close the control loop;
• data collection and policy training, the hottest topic in robotics right now.

My strong opinion: robots like this will teach more people to program physical things than any course. Not because they're perfect, but precisely because they're imperfect in a visible way. The cable stretches, the gripper misses by two centimeters, and you have to figure out why. That's real engineering. Very different from a green unit test.

And an honest warning: $800 is the price of the kit, not of your patience. Set aside one weekend to install the anchors and another to calibrate. Your landlord might also have an opinion about holes in the ceiling.

What it all comes down to

Stringman won't replace an industrial arm, and it doesn't want to. Its merit is something else: proving you can reach an entire room with cheap motors, wires and good math, and leaving it all open for anyone who wants to improve it.

For those of us who live on TypeScript and databases, it's a good reminder. Often the expensive solution is the obvious one, and the cheap one just needs a change of perspective. Instead of building a stronger arm, hang the gripper. The same goes for code.

If you like watching simple ideas turn into things that work, take a look at the projects I've been building.

The camera that flies over the field during football broadcasts now fits in your living room for $800.

Stringman is an open source robot with no arm at all. It's just cables tied to the corners of the room, pulling a gripper wherever the math says.

A robotic arm with that kind of reach costs as much as a car. Here the problem moves somewhere else: the motors stay fixed on the wall and only the gripper moves.

And the inverse kinematics? It's Pythagoras in 3D. The hard part is the real world: cables that stretch, spools that change diameter, a gripper that misses by two centimeters.

That's why I believe robots like this will teach more devs to program physical things than any course. Often the expensive solution is the obvious one, and the cheap one just needs a different point of view.

I wrote about it on the blog. Would you build one of these at home?

#Robotics #OpenSource #Hardware #Programming #ArtificialIntelligence