This section explains how Apt came about, and why it was born so early in the history of computing.
After World War II, the US began a phase of military buildup as a result of the Cold War. A key arena was the aerospace industry, for the obvious military importance of aircraft and intercontinental ballistic missiles. Maintaining superiority in these weapons was critical to Cold War strategy. It was early in this period that much technological progress was made in machining.
The aircraft after World War II put increasing demands on the manufacturing techniques of the time. The construction of jet and supersonic aircraft put new demands for high structural integrity with light weight. For example, for increased strength with less weight, critical structural parts were being machined in one piece instead of assembled from several components. This greatly increased the complexity of parts and made for long lead times and scrapped parts. Many parts with complicated geometry were created with the use of template controlled machine tools. As the name implies, these machine tools were guided by a tracer following a master part. The disadvantage with such a scheme is that the setup of such a machine tool is dependent on the creation of a master part, which is necessarily time-consuming and costly, and the capabilities are somewhat limited. The accuracy of such template-controlled tools was also reaching the limits of the precision needed.
The Air Force closely followed the needs of the industry in their desire to get their products on time and within budget. Consequently, they sponsored research projects that were deemed potentially useful for solving the problems that the aerospace industry was facing. One such project that they wisely funded was the NC machine tool, an idea based on ideas by an industry manufacturer named John Parsons, and fully developed by MIT.
John Parsons was the president of the Parsons Corporation, an aerospace company. He had served an apprenticeship in tool and die work, and this particular apprenticeship had given him some ideas on different ways of attacking some of the problems that were beginning to hamper the companies at this time. It was a common occurance in machine shops of the time, when faced with the obstacle of machining an irregular contour, to first lay out the profile on the workpiece, then to make closely spaced, drilled holes along this contour. The unneeded material could then be punched out. This left ridged material that would need to be filed or otherwise smoothed. When faced with a similar problem in producing templates for helicopter airfoils, only one where the accuracy was much more important, he remembered this technique and realized that if the holes could be spaced closer together, filing would be reduced and accuracy increased. A Parsons employee named Frank Stulen had great success in applying this idea by using a mechanical calculator to calculate the multitude of points required to position the tool for creating the template. When a contract for a complicated, 3d part was submitted for bid, Parsons realized the same concepts used in the 2d airfoil templates could be applied in the 3d case as well. The only remaining problem was how to realize this concept, since the number of discrete points was huge, and most likely beyond the capabilities of a human operator to make such positionings quickly and without error.
MIT played a large role in the shape of the NC technology that ultimately came about. John Parsons can be said to have driven the initial work in NC, and was responsible for getting the Air Force interested in the concept, but until MIT became involved, the project was not nearly so broad and well-developed. Parsons' ideas centered around discrete positioning of a machine tool to rough the shapes needed in his specific application by reading punch cards. This was more or less the automation of a time consuming manual positioning idea. MIT studied the problem in-depth and realized it's much broader potential to automate the entire machining process, and in particular, drive the tool in a continuous, 3d path in space. In 1952, they had succeeded in creating a demonstration machine with these concepts.
The new technology was a hard sell. To many in the industry, the MIT machine tool ultimately designed was far too complex to be useful in an environment like a machine shop. Computers were a very new concept, and the layman regarded these machines as mysterious and unfathomable. It didn't help that the first machines used vacuum tubes which were prone to burn out. Finding a capable technician to repair the machines was another problem.
During the period when MIT was developing the NC control concept, they were careful to to look at the project on all levels, and the economics of machining this way compared to current methods were important if the NC concept were to develop roots. They found that they could be competitive with machine shops on some products, but the make-ready costs were a major obstacle. The main cost was the extremely time-consuming process of doing the mathematical calculations required to create the control tape. To give some idea of the task facing the tool programmer, consider that the early controls required a series of straight-line moves to generate curved surfaces. To keep the surface relatively smooth, many line segments would need to be generated, and these would have to be calculated using a mechanical calculator. Clearly, this task was an economic drain.
With this background, MIT began a program, with industry support and Air Force Funding, to develop what became Apt. An MIT graduate student named Doug Ross was given the task to put together a system and coordinate the various outside efforts. The final result was a well thought out system that was revolutionary for its day.
For a detailed exposition on the history presented here, see the book
Numerical Control: Making a New Technology
By J Francis Reintjes
Oxford University Press