A title that needs a caveat
There was no single “first” computer-aided manufacturing program. The term CAM grew out of several overlapping efforts in the 1950s and 1960s to use computers to prepare instructions for machine tools, and universities, aircraft makers, computer companies and government agencies all contributed. What follows describes the main strands and says where credit is shared.
The problem: cutting complex shapes
Numerical control, usually shortened to NC, means directing a machine tool with coded numbers instead of a human hand on the controls. Early work in the late 1940s and early 1950s was driven largely by the aircraft industry, which needed parts with complex curved shapes, such as wing components and propeller or rotor blades, that were slow and costly to make by conventional methods.
Several groups are associated with this period. One inventor and his company, working with Air Force support, explored machining guided by punched cards of calculated positions. A laboratory at the Massachusetts Institute of Technology, which had experience with servomechanisms, demonstrated a numerically controlled milling machine in the early 1950s. Accounts of who deserves the most credit differ, and it is fair to say both contributed to the field.
Why programming was a bottleneck
Early NC machines needed very detailed instructions: every position the cutting tool should move to, in sequence. For a simple drilled plate, a person could write these by hand. For a curved surface, the path involves many calculations, and even a small arithmetic error could ruin an expensive piece of metal.
This is where computers entered. A mathematician or programmer could describe the shape, and a computer could calculate the positions the tool needed to pass through. The work resembled other scientific programming of the era on mainframe computers, but its output was unusual: a punched paper tape or card deck to be carried to the shop floor.
APT and the language approach
The best-known response was a language, not a one-off program, much as early software tools of the 1950s let people describe tasks in words rather than machine code. APT, usually expanded as Automatically Programmed Tool, was developed at MIT in the second half of the 1950s, with Douglas Ross prominent in the effort. A programmer wrote statements in a vocabulary resembling geometry, defining points, lines, circles and planes, and then describing the path the tool should follow along them. The computer worked out the detailed tool movements.
APT did not stay a single laboratory’s project. A group of companies, working through an industry organization, contributed to its growth in the early 1960s, and the language was used for many years, especially in aerospace. Many versions and descendants appeared.
Other early systems
APT is not the whole story. Other groups built their own programming systems, some simpler and aimed at machines and computers that were more widely available.
- Computer manufacturers produced languages that could run on their own machines, so smaller shops could use them without a very large computer.
- Researchers and companies in several countries, including Germany, developed languages for particular classes of work, such as drilling and simple milling, which were easier to learn than full APT.
- An engineer at a large electrical company is often credited with early work on systems that connected geometry to machining. That work later became entangled with the beginnings of early CAD software.
Because these efforts overlapped, and because records are scattered across company reports and conference papers, histories do not always agree on dates or relative importance. Treat precise claims of priority with caution.
The post-processor
One idea that proved lasting was the split between general geometry and machine-specific details. The main program calculated the path of the cutting tool in a general form. A separate piece of software, a post-processor, translated that path into the exact codes and format a particular machine and control unit expected. This division meant the same part description could be adapted for different machines by changing only the post-processor.
Post-processors remain part of manufacturing software. The need for them partly reflects the lack of a single universal code, even though standards for machine code appeared in the 1960s and were widely used.
From tape to shared files
In the 1960s, the usual workflow was long and indirect. An engineer produced a drawing, a part programmer interpreted the drawing, wrote a part program, ran it through the computer, and collected a tape. The tape was then tested on the machine, often with a trial cut in soft material, and then corrected. The connection with design was by paper, not by shared data.
That began to change when design systems could store geometry that manufacturing software could read directly. Later, microprocessors put the control computer inside the machine itself, which extended the idea further.
What the early work achieved
The early CAM programs did not make factories automatic. Operators still set up machines, loaded materials, checked parts and fixed problems. What the languages did was reduce the arithmetic burden and open the door to shapes that had been impractical to produce by hand. They also created a new specialty, the part programmer, whose skills combined machining knowledge with a new kind of computer language.
It is also worth being careful about economic claims. Within the wider story of industrial automation, numerical control spread slowly, because machines and programming effort were expensive, and the benefits depended on the kind of work. Large aerospace and defense contractors were early adopters partly because they had both the complex parts and the budgets. Smaller shops tended to wait for cheaper equipment.