Magic Trick Robot: From Automata to ABB Precision

A magic trick robot is an old idea: from Maillardet and Robert-Houdin to a 2026 card-swapping robot hand, and where ABB YuMi and GoFa arms fit in today.

INDUSTRIAL ROBOTICS

Chat With Robot

10/19/20265 min read

A magic trick robot sounds like a novelty, but the idea is old. Long before electronics, clockmakers built figures that wrote, drew and even performed conjuring tricks, and audiences paid to wonder how they worked. Today's robot hands can swap a card or vanish a coin, and the skills behind those tricks, steady fingers and exact timing, are the ones factories need too. This post follows the story from brass automata to a recent Chinese robot hand doing card tricks, then looks at where ABB's YuMi, GoFa and IRB 1200 robots fit.

A hand holding a playing card flat against the palm
A hand holding a playing card flat against the palm

A playing card held palmed in a hand, the kind of sleight a robot hand now copies. Photo: Mlee1223 / Wikimedia Commons (CC BY-SA 4.0)

How clockmakers built the first magic machines

Henri Maillardet, a Swiss mechanician working in London around 1800, built an automaton that writes poems and draws pictures. Its cams store the movements, so the brass boy can repeat two French poems, one English poem and four drawings, including a ship and a Chinese temple. The Franklin Institute in Philadelphia received the broken machine in November 1928, and Andrew Baron restored it in 2007. A program does not need electronics. It only needs something that remembers a motion, and a set of cams does that. The same idea, a stored sequence replayed exactly, still sits at the heart of every robot program today.

A mechanical boy figure at a small desk inside a glass display case
A mechanical boy figure at a small desk inside a glass display case

The Maillardet automaton at the Franklin Institute, a cam-driven figure that writes and draws. Photo: Daderot / Wikimedia Commons (public domain)

Pierre Jaquet-Droz and his team in Neuchâtel made three famous figures in the 1770s: the Writer, the Draughtsman and the Musician. The Writer dips a quill and pens a line of text while its eyes follow the pen, and the Musician plays a small organ with its fingers. Audiences took these as marvels, and for a stage magician the lesson was plain: a machine that seems to think draws a crowd. Building one took the skills of a watchmaker, which is why so many early automata came from clockmakers rather than from theatres.

A doll-like automaton figure behind a wooden organ case with a red cloth front
A doll-like automaton figure behind a wooden organ case with a red cloth front

The Jaquet-Droz Musician, built in the 1770s, which plays a small organ. Photo: Gre regiment / Wikimedia Commons (CC BY-SA 4.0)

Jean-Eugène Robert-Houdin, born in Blois in 1805, trained as a watchmaker and then turned to the stage. He built automata, among them a figure that performed the cups and balls, and a writing and drawing figure that he showed to King Louis Philippe I and later sold to P. T. Barnum. In 1845 he opened his own 200-seat theatre in the Palais Royal in Paris. He is often called the father of modern magic, and his act drew on clockwork and on the newly discovered electromagnetism.

Black and white portrait of an elderly man in a dark coat with a bow tie
Black and white portrait of an elderly man in a dark coat with a bow tie

Jean-Eugene Robert-Houdin, the watchmaker turned stage magician, in a 19th century portrait. Photo: Unknown author via Wikimedia Commons (public domain)

Robot hands, card tricks and ABB precision

The modern version of the idea showed up in 2019, when German iPad magician Simon Pierro bought a SoftBank Pepper robot and began training it as an assistant for a new act called Artistic Intelligence. In one video he uses a magnet to pull an iPhone out of the screen on Pepper's chest. Pepper plays the stage partner there, and the human still does the tricks. The act pairs a very modern machine with a very old showman's instinct. That shows where the hard part sits: getting a machine to move its fingers well enough to fool a close audience.

A white humanoid Pepper robot with a tablet screen on its chest
A white humanoid Pepper robot with a tablet screen on its chest

A SoftBank Pepper robot, the model magician Simon Pierro trained as a stage assistant. Photo: SoftBank Robotics Europe / Wikimedia Commons (CC BY-SA 4.0)

That hard part is now being tackled directly. On 5 October 2026, Interesting Engineering reported that LinkerBot, a Beijing startup founded in 2023, had released a video of its LinkerHand O30 doing a card change, pulling a toy car out of a thin card and making a coin vanish. The company says the hand has 20 active degrees of freedom, weighs 730 grams, takes commands 500 times a second and repeats positions to within 0.2 mm. Those figures are LinkerBot's own claims, not independent tests, and the hand is pitched at electronics, automotive and aerospace work as much as at stages.

A brown robot arm with a metal gripper holding a cast metal part
A brown robot arm with a metal gripper holding a cast metal part

A robot gripper holding a cast part at NIST's dexterous manipulation testbed. Photo: NIST (public domain)

ABB does not sell a magician, and I found no source saying it plans to. Its robots do have what a trick needs, though. The dual-arm YuMi IRB 14000 handles small parts at human scale, and the single-arm YuMi IRB 14050 and the GoFa CRB 15000 can be taught by hand, so a designer could show a motion once and let the arm repeat it. ABB's YuMi even conducted an orchestra in Pisa in 2017, which shows that smooth, timed motion can hold a live audience. Our earlier posts on robot performers and the dance robot cover that side.

A dual-arm YuMi robot handling small parts on a bench while a technician watches
A dual-arm YuMi robot handling small parts on a bench while a technician watches

ABB's dual-arm YuMi assembling small parts beside a lab worker. Photo: ABB

What comes next for robot magicians

A card trick robot at home looks possible within a few years, but probably as a small arm that does one routine very well, such as a coin vanish or a card reveal at a child's birthday party, rather than a hand that does everything. The open problems are cost, reliability and fooling people at arm's length, because a magic audience sits close and watches the hands. A trick that works from only one camera angle is a demo, and a stage act needs the thousandth performance to match the first.

A YuMi robot on a podium in front of an orchestra beside a singer
A YuMi robot on a podium in front of an orchestra beside a singer

ABB's YuMi conducting the Lucca Philharmonic with Andrea Bocelli in Pisa, 2017. Photo: ABB

In factories the payoff is less theatrical. The finger control that slides a card out of a deck is the same control that electronics, medical devices and watch parts need. Dexterous hands will probably arrive first as grippers on existing arms, which is how LinkerBot pitches its own. Safety matters too: a hand that the company says can carry 30 kg needs guarding or force limits near people. ABB's SafeMove software and its cobot designs already handle this for arms, and the same approach would apply to a new hand.

A woman moving a grey GoFa cobot arm by hand at a work table
A woman moving a grey GoFa cobot arm by hand at a work table

Teaching an ABB GoFa arm a motion by guiding it by hand. Photo: ABB

Skills are the third problem. Teaching a robot a trick takes someone who understands both the sleight and the motion program. Engineers can build and test such programs in RobotStudio before touching a real arm, and ABB's Wizard easy programming lowers the bar for people who are not programmers. A hobbyist with a small arm, a pack of cards and a few weeks of practice may get further than a lab did ten years ago.

A compact white ABB IRB 1200 robot arm on a red base
A compact white ABB IRB 1200 robot arm on a red base

The new generation ABB IRB 1200, a compact six axis arm for fast small part work. Photo: ABB

Innovation

AI solutions for effortless ABB robot control.

Automation

Robotics

ceojohntran@chatwithrobot.net

+84905311611

© 2025. All rights reserved.

qtran1215@gmail.com