Industrial robot

Based on Wikipedia: Industrial robot

In 1937, a British engineer named Bill Griffith P. Taylor built a robot out of Meccano toy parts. It ran on a single electric motor, read instructions from punched paper tape, and could stack wooden blocks in programmed patterns. To figure out how far each motor needed to turn, Taylor plotted the movements on graph paper first, then punched the data onto tape by hand.

That crane-like contraption, assembled from children's construction toys and powered by solenoids clicking on and off, was the earliest known industrial robot.

Today, over four million industrial robots operate in factories worldwide. They weld car frames with superhuman precision. They paint surfaces without leaving a single drip. They pick up circuit boards and place components at speeds no human hand could match. The journey from Meccano blocks to modern manufacturing took less than a century—and the acceleration shows no signs of stopping.

What Makes a Robot "Industrial"

An industrial robot is, at its core, a programmable machine that can move in three or more directions. The word "axes" describes these directions of movement. Think of it this way: if you can only move left and right, you have one axis. Add forward and backward, and you have two. Add up and down, and you have three. Three axes let you reach any point in a room.

But reaching a point isn't enough. You also need to control how your tool arrives at that point. Imagine trying to insert a screwdriver into a screw—you need to approach at exactly the right angle. This requires three more axes of rotation, often called yaw, pitch, and roll, the same terms pilots use to describe an airplane's orientation.

Six axes total. That's what it takes for a robot arm to reach any position, at any angle, in three-dimensional space.

The Six Species of Industrial Robot

Not all robots need all six axes, and different tasks call for different body plans. Over the decades, engineers have converged on six main designs, each with its own personality.

Articulated Robots: The Human Mimics

These are the robots you probably picture when you hear the word. They look like mechanical arms, with joints that bend like shoulders, elbows, and wrists. The resemblance to human anatomy isn't accidental—it gives them extraordinary flexibility. An articulated robot can reach around obstacles, fold itself into compact spaces, and approach a workpiece from almost any direction.

They're the most common type in factories today.

Cartesian Robots: The Straight Shooters

Named after René Descartes, the mathematician who invented coordinate geometry, these robots move only in straight lines—left-right, forward-backward, up-down. No graceful sweeping arcs here. Picture a 3D printer's print head moving along rails, or a crane sliding along a gantry.

What they lack in elegance, they make up for in simplicity. Straight-line motion is easier to program, easier to predict, and easier to make precise. If your task involves moving objects between fixed points on a grid, a Cartesian robot will do it reliably and cheaply.

Cylindrical Robots: The Rotating Towers

Imagine a tower that can spin around its base and extend its arm outward while also moving up and down. That's a cylindrical robot. The name comes from the shape of its working envelope—the region of space it can reach—which forms a cylinder around its base.

These compact machines excel at reaching into tight workspaces. They were among the earliest designs used in manufacturing, and they're still found in applications where space is at a premium.

Spherical Robots: The Pioneers

The spherical robot was one of the first types ever used in industry. Instead of sliding along tracks, every joint rotates. The working envelope forms a partial sphere around the base. These robots found their first homes in die-casting plants and injection molding facilities, tending machines that shaped molten metal and plastic.

They've been largely superseded by articulated robots, but they hold an important place in the history of automation.

SCARA Robots: The Speed Specialists

SCARA stands for Selective Compliance Assembly Robot Arm—a mouthful that describes a very specific talent. These robots have two parallel joints that allow movement in a horizontal plane, like a human arm reaching across a table. A vertical shaft at the end can plunge up and down.

The magic word here is "compliance." SCARA robots are designed to be rigid vertically but slightly flexible horizontally. This makes them perfect for assembly tasks where parts need to be pressed together. If there's a tiny misalignment, the arm can adjust slightly rather than jamming or breaking something. Watch a SCARA robot insert electronic components onto a circuit board, and you'll see why they dominate precision assembly lines.

Delta Robots: The Lightning Hands

Delta robots look nothing like the others. Instead of a single arm, they use three or four lightweight arms arranged in a triangle or square, all connected to a small platform at the bottom. The arms move in parallel, controlling the platform's position with extraordinary speed and precision.

These are the robots you see in food packaging plants, picking up chocolates or cookies and placing them into boxes at dizzying speeds. They can make hundreds of picks per minute. The secret is their low mass—because the motors sit at the top and only the lightweight arms move, there's very little inertia to overcome. Delta robots can change direction almost instantaneously.

Serial Versus Parallel: Two Philosophies of Movement

Beyond the six body types lies a deeper architectural divide. Most robots are "serial"—their joints connect end to end in a chain, like links in a necklace. The shoulder connects to the upper arm, which connects to the elbow, which connects to the forearm, and so on to the wrist and gripper.

This design is intuitive and flexible, but it has a flaw. Each joint's imprecision adds to the next. If your shoulder is off by a tiny amount, your elbow amplifies that error, your wrist amplifies it further, and by the time you reach your fingertip, you might be millimeters away from where you intended.

Parallel robots take a different approach. Multiple chains connect the base to the end platform simultaneously. Any error in one chain gets averaged out by the others. The Delta robot is the most famous example—its three arms work together, each one constraining and correcting the others.

The tradeoff? Parallel robots have more limited ranges of motion. What they lose in flexibility, they gain in precision and rigidity.

The Measure of a Machine

Engineers evaluate robots on several key specifications, and two of them are often confused: accuracy and repeatability.

Accuracy is how close a robot can get to a commanded position. If you tell it to move to a specific point in space, how far off will it actually be? Accuracy depends on how well the robot's internal model matches reality—the exact length of each arm segment, the precise relationship between motor turns and joint angles.

Repeatability is different. It measures how consistently a robot returns to the same position. A robot might never hit exactly the spot you asked for—maybe it always ends up one millimeter to the left—but if it lands in that same wrong spot every single time, its repeatability is excellent.

Here's the counterintuitive truth: repeatability matters more than accuracy for most industrial applications.

Why? Because you can teach a repeatable robot where to go. You physically guide it to the right position, record that position, and the robot will return there faithfully forever after. The exact coordinates don't matter as long as the robot hits the same mark every time. Modern industrial robots routinely achieve repeatability of a tenth of a millimeter—about the thickness of a human hair.

Accuracy, by contrast, is what you need when you can't teach the position in advance. If a robot has to calculate where to go based on external measurements—maybe it's reading a barcode to figure out which product to pick—then it needs accurate internal models of its own geometry.

The Race to Automate

George Devol filed the first robotics patent in 1954. Two years later, he partnered with Joseph Engelberger to found Unimation, the world's first robotics company. Their early machines were called "programmable transfer machines" because that's mostly what they did—transfer objects from one spot to another. They ran on hydraulic power and used a teaching method that's still common today: an operator would physically move the arm through the desired motion while the robot recorded the joint angles, then play them back during operation.

For years, Unimation's only American competitor was Cincinnati Milacron. The industry remained small and specialized.

Then came the 1970s.

In 1969, a Stanford researcher named Victor Scheinman invented something revolutionary: the Stanford Arm. Unlike previous robots, it was entirely electric—no hydraulics—and it had six fully articulated joints. More importantly, it could follow arbitrary paths through space, not just move between fixed points. This opened the door to welding, where the robot must trace precise seams, and assembly, where complex motions are the norm.

Scheinman later sold his designs to Unimation, which developed them into the PUMA—the Programmable Universal Machine for Assembly. The PUMA became one of the most influential robots in history.

Meanwhile, Europe and Japan were racing ahead. In 1973, both ABB Robotics in Sweden and KUKA Robotics in Germany introduced their first commercial robots. ABB's IRB 6 was among the first microprocessor-controlled robots available for purchase. The first two units went to a Swedish company to grind and polish pipe bends—not glamorous work, but the kind of repetitive, precise task that robots excel at.

The late 1970s brought a robot boom. General Electric, General Motors, and dozens of startups piled into the market. Japan's FANUC formed a joint venture with General Motors. At the peak of the boom in 1984, Westinghouse bought Unimation for 107 million dollars.

The shake-out was brutal. Most American companies failed. The survivors were mostly European and Japanese: ABB, KUKA, FANUC, Kawasaki, Yaskawa. Today, these giants still dominate the global robot market.

The Autonomous Frontier

The robots we've discussed so far are, in a sense, sophisticated playback machines. They repeat recorded motions with inhuman precision, but they don't think. They don't adapt to unexpected situations. They can't identify objects or make decisions.

That's changing.

Modern industrial robots increasingly incorporate machine vision—cameras and software that let them see and interpret their environment. A vision-equipped robot can find a part that's been placed slightly askew, recognize different product types on a conveyor belt, or detect defects that human inspectors might miss.

Artificial intelligence is the next frontier. Robots are beginning to learn from experience, adapting their movements to achieve better results. They're starting to work alongside humans without safety cages, sensing when a person is nearby and adjusting their speed or stopping entirely.

Interestingly, the concept of autonomous robots predates the industrial robot by a decade. In the late 1940s, a British neurophysiologist named W. Grey Walter built two small robots he called Elmer and Elsie. Shaped like tortoises, these machines had no programming in the traditional sense. Instead, their simple electronic brains responded to light and touch, creating surprisingly complex behaviors. They would seek out light sources, avoid obstacles, and even recognize their own reflections. Walter built them to study how simple rules could produce seemingly intelligent behavior.

Elmer and Elsie didn't do useful work, but they demonstrated something profound: machines could respond to their environment in ways that looked almost alive. Seventy years later, their descendants are learning to see, to reason, and to adapt—while also welding car bodies at a rate of thousands per day.

Why This Matters Now

As of 2023, over four million industrial robots are working in factories around the world. That number has roughly doubled in the past decade. The robots are getting cheaper, smarter, and more versatile. Tasks that once required expensive custom automation can now be handled by general-purpose robot arms that cost less than a year's salary for a human worker.

The implications ripple through the global economy. Supply chains are being reshaped by the promise—and the reality—of flexible automation. Companies that once moved manufacturing overseas for cheap labor are reconsidering. A robot costs the same whether it's in Shanghai or Stuttgart or South Carolina.

From Bill Taylor's Meccano crane to the four million mechanical workers in today's factories, industrial robots have traveled an extraordinary path. The punched paper tape has given way to neural networks. The hydraulic cylinders have been replaced by precision servomotors. The robots have gained eyes, and they're beginning to gain something like understanding.

What remains constant is the fundamental bargain: trade flexibility for precision, and repetitive human labor for tireless mechanical consistency. Every arc welded, every component placed, every painted surface represents a negotiation between human intent and robotic capability.

The negotiation is far from over.