The earliest known industrial robot was built from Meccano parts
In 1937 an engineer named Bill Griffith P. Taylor assembled a crane almost entirely from Meccano, the toy construction set, and drove it with one electric motor. A punched paper tape told it what to do, and it could stack wooden blocks in set patterns. It meets the formal definition of an industrial robot.
Taylor's machine, published in Meccano Magazine in March 1938, moved on five axes, including gripping and rotating its grab. He plotted the motor revolutions needed for each move on graph paper, then punched them into tape that energised solenoids to throw the crane's levers. A complete replica was built in 1997. The idea it demonstrated, a machine that is programmable, automated and able to move on three or more axes, is still how industrial robots are defined.
The industry proper began with George Devol, who applied for robotics patents in 1954, and the company Unimation, founded with Joseph Engelberger in 1956. Its hydraulic arms were first used to shift objects a few feet from one place to another; operators taught them by recording joint angles, which the machines then replayed. In 1969 Victor Scheinman at Stanford built an all-electric six-axis arm that could follow arbitrary paths, opening the way to welding and assembly. Europe moved fast: ABB and KUKA both launched robots in 1973, and the first two ABB units went to a Swedish firm to grind and polish pipe bends.
A boom followed in the late 1970s as Japanese conglomerates piled in. Westinghouse bought Unimation in 1984 for 107 million dollars, but only a handful of non-Japanese makers survived. Today robots weld, paint, assemble, inspect and pack, and the International Federation of Robotics estimated about 4.66 million were working worldwide in 2024.
Shapes vary with the job. Articulated arms, resembling a human arm, are most common. SCARA robots excel at precise sideways assembly moves, and delta robots, built from parallel linkages, are prized for rapid pick-and-place. Parallel designs are stiffer because errors in each short chain average out instead of piling up, as they do along a serial arm. Increasingly, machine vision and artificial intelligence let robots handle parts that are not perfectly positioned.
Source: Industrial robot
