Robot arm

A palletising robot made from flat steel. Three linear actuators drive it, and nothing slides. Every joint turns.

Context: SDU · Integrated Design 3 · 2025
Built from: 1.5 mm steel · rivets · axles · 3D print

The finished robot arm in the workshop: dark steel links with orange rivet caps, a gripper with white 3D-printed antlers and a glowing red ball for a nose, above the drop-off platform. — Final workshop, week 50It was December. It got antlers.

01 — The brief

Move two parts from a low platform to a high one. As fast as you can.

We were handed a base that turns about 80°, driven by a linear actuator. The rest was ours: two more degrees of freedom, each on its own linear actuator, and a pneumatic gripper. All in laser-cut steel, bent by hand. On the last day, every group raced the two parts against the clock.

02 — The idea

An actuator pushes in a straight line. We wanted every joint to turn.

So no actuator slides a part along. Each one pushes on a short lever behind the pivot, so a short stroke turns into a long swing at the tip. That gives the arm its reach. One actuator, one joint. The arm moves like an arm.

The gripper has to face the parts the same way wherever the arm is. Two parallel links do that on their own. No extra motor, nothing to control.

Fig. 01 — Push behind the pivot, swing at the tip

Fig. 02 — The arm moves, the gripper doesn’t tilt

03 — Build first

No report to write. So we built, and tried, and built again.

We found the geometry with cardboard and LEGO, not calculations. The cardboard arm went straight onto the real base, next to the real platforms. If it couldn’t reach both, it was wrong, and we could see it in seconds.

A full-size cardboard model of the arm, with pin joints and cardboard tubes standing in for actuators, mounted on the real rotating base next to the steel drop-off platform. — Cardboard at full scale, on the real base

CAD model of the whole setup: perforated table, the arm on its rotating base, the high drop-off platform and the two parts on the low pickup plate. — The full setup in CAD — base, arm, two platforms, two parts

04 — Flat first

Every part starts as a flat sheet.

The laser cuts it, we bend it by hand, rivets hold it. So we drew the arm as sheet metal from the start. The first version was flat plates full of oval holes. The built arm folds each link into a closed profile: the same sheet, much stiffer.

Early CAD of the arm, three-quarter view: flat link plates with oval cut-outs and triangular joint plates.

Early CAD of the arm, side view, with the two linear actuators visible inside the lower frame.

Early CAD — both actuators sit low, inside the frame

05 — Built

Laser-cut steel, pop rivets, axles and 3D prints. Nothing else.

1.5 mm steel, cut and bent in the SDU workshop. Rivets join the sheets, steel axles carry the joints, and 3D prints fill in where sheet metal isn’t the right answer.

06 — Driven by hand

Three actuators. Three controls. One person at the box.

There’s no program. A control box drives each actuator on its own, and the operator plans the path in their head: down, grip, up, swing, down, release. Because each control turns one joint, what you do with your hand is what you see the arm do.

The large part, low platform to high

07 — What I learned

The best advice came from the workshop, not the screen.

SDU has machinists in the workshop with years of experience in exactly this: cutting, bending and joining steel. Talking to them early saved us from mistakes we would only have found at the bending machine. Now I go to the people who make things before I commit to a design.

People: Use the experience around you. A five-minute question to a skilled craftsperson beats a day of guessing.
CAD: Drawing in sheet metal from the start, so what’s on screen is what the laser cuts and the machine bends.
Rapid prototyping: Cardboard and LEGO answer geometry questions in minutes. Build it, hold it up to the real thing, change it.

Next projectGOK GOK →