I had a PDP-8 years ago that was used for controlling a spherical photometer for a manufacturer of lighting systems. Seems a good match to the capabilities to use for machine control.
PDP-11's saw a great deal of use as machine controllers for a number of reasons, the most important of which was the ability to do hard-realtime control. I've seen several at this point, including an 11/60 that was used to control a specialized scanner that was built to scan glass photographic plates such as produced by telescopes prior to advent of the CCD astronomical imager.
11/23's were very popular for this usage, since they were fast enough yet relatively inexpensive enough for this task. Although I did see one setup of two scanners, based on Perkin-Elmer PDS 2020 microdensitometers, that used VAXstations, and interfaced to the realtime hardware's subprocessors via a SCSI-connected CAMAC crate. The realtime positioner servo was an HP laser interferometer using several 68020s to do the realtime job, positioning the scanning spot to a precision of 63nm. Hardware-clocked FIFOs produced the position-stamped samples, since the 90MHz VS4000 was simply neither fast enough nor realtime enough running VMS to handle it.
11's also got a lot of usage for telescope controllers, again mostly thanks to the hard-realtime capabilities but also due to the great language support of the day on RT11 and RSX11, mostly FORTRAN for scientific uses but later on C took a foothold. Much astronomical software used professionally (such as IRAF) still has a significant FORTRAN heritage.
Today's closest equivalent wouldn't be the RasPI or similar Linux SBCs, but the ATMega328-series SBC's (aka Arduino), which have similar speeds to the various PDP-11's (ATMega 328 SBC's typically clock at 14.5MHz or thereabouts and are quite efficient; high-end PDP-11's such as the 11/93 clock in the same basic neighborhood for the microcycle timing (18MHz) which yields an effective clock (for a direct comparison) of 4.5MHz. The 11 being 16-bit versus the ATMega being 8-bit can make the 11 a bit faster clock-for-clock, but I personally would find a direct comparison (maybe the PI benchmark referenced in the thread at
http://www.vcfed.org/forum/showthread.php?69070-VAX-7000-640-CPU-frequency would be interesting). I/O robustness is totally irrelevant to machine control purposes, for the most part, so even though the 11's can likely run rings around the ATMega328 boards in terms of disk I/O etc it just doesn't matter for that application.
RasPI and similar SBCs aren't really built to be hard-realtime, although there are exceptions.