This article, which is published under the heading “Artificial Intelligence,” is the kind of pop-scientific survey of this interesting topic that we can expect to find in BYTE magazine.

The flavor of the article is indicated in the first paragraph. The authors state flatly that current machine systems are “binary . . . not the full gray-scale range provided by an input device such as a TV camera. . . . Also . . . the systems . . . will not tolerate events such as parts obscuring one another or lying propped up at odd angles. A seemingly simple but elusive industrial vision task, beyond the capability of all current systems, is the ‘bin-picking’ problem.”

In fact, there are systems on the market that use four or more bits of gray scale. There are systems that pick objects jumbled together in a bin, and that necessarily analyze occluding parts.

The tone is strangely reminiscent of the “bionics” approach of the 1960s. Sometimes the approach is stood on its head: “many signs point to biological versions of the computer vision hierarchy.” In a moment of boldness, the authors even surmise that, “it seems likely that the biological system has also [sic] solved the parallel-computation problem that is emerging as the issue of formal vision studies.”

For the BYTE audience, this is probably a stimulating introduction to an important field of contemporary research. Still, one has to wonder why DARPA and NSF supported its preparation.