What Made It an Arm
A robot arm is a chain of rigid links connected by joints, with a gripper or tool at the end. Each joint can be driven separately, so the end of the arm can be placed at many positions and angles within its reach. That sounds obvious today, but in the 1950s the combination was new in commercial machines: a human-arm-like structure, powered by its own drive system, controlled by a stored program instead of a person at the controls.
Earlier machines had some of these elements. Remote manipulators used in the nuclear industry in the 1940s let an operator handle radioactive material from behind a barrier, and they copied the operator’s hand motions. But they were controlled by a person in real time. The step that led to the industrial robot arm was putting the memory of the motion inside the machine.
The Hydraulic Drive
The early Unimate arm used hydraulics, meaning pressurized oil moving pistons and motors. In the early days, this was a practical choice because hydraulic actuators deliver strong force from a compact unit and can move heavy parts smoothly. The arm was built to lift and place heavy, hot items, which suited this strength. The trade-offs were familiar to anyone who has seen hydraulic equipment: pumps and hoses were needed, oil could leak, and precision was limited compared with later designs.
By the late 1960s and 1970s, designers increasingly turned to electric motors, which were cleaner and easier to control accurately as electronics improved. A widely cited example was a small electric arm built by Victor Scheinman at Stanford in about 1969, and later commercial arms followed the same path. Today hydraulic arms remain in some heavy applications, but most industrial arms are electric.
Programmed Memory
The signature feature was not the arm but the memory. Rather than writing code, an operator would typically lead the arm through the desired motion step by step, and the controller stored each position. Early accounts describe this storage as being on a magnetic drum, a rotating cylinder that held data as magnetic patterns, which was an early form of computer memory. During work, the controller replayed the stored positions in order, so the arm repeated the same sequence over and over. This “teach and playback” approach meant that factory staff did not need to be programmers.
The control was limited. The arm followed a recorded path and did not use cameras or touch sensing to check what it was doing. If a part arrived in the wrong place, the arm would still go through its motions. For this reason, early robot installations needed parts to be presented reliably, with fixtures and conveyors doing the positioning. Later machines gained more flexible control as electronic computers became small and cheap enough to be embedded, and sensors were added over time.
The Jobs It Did
Tasks that suited this kind of arm were dangerous, dirty or dull, and repetitive. The first well-known use was removing hot castings from a die-casting machine and stacking or cooling them, a job with heat and fumes. Welding followed: in the 1960s, spot welding of car bodies became a major application, because the arm could hold a heavy welding gun and repeat the same weld points consistently. Other early applications included loading and unloading machines and handling parts in forging and press operations. Painting and sealing came into use as well, partly because they removed workers from fumes.
Such tasks fit into larger systems, such as assembly line production, where the arm’s consistency helped the whole sequence. The cost and complexity of installing arms, however, meant that adoption was gradual and concentrated in large manufacturers, chiefly in the car industry.
How the Design Spread
The basic form of the jointed arm has proved durable. Modern industrial robots commonly have six axes, which lets the tool reach a point from different angles, and many designs are close descendants of the structure that developed in the 1960s and 1970s. The same arm configuration appears in laboratories, in packaging lines, and in a modified form in surgical robots, where very precise motion matters. Researchers working on soft robotics, by contrast, explore alternatives to rigid links.
It is fair to add that the credit for the first robot arm is shared. Devol’s patent ideas, Engelberger’s business effort and the engineers who built the machines all mattered, and other inventors elsewhere were working on related ideas around the same time. What the Unimate arm showed, rather than a single technical trick, was that a programmable arm could be a reliable factory tool.