Erm... What?

What matters to us is that in 1982 NASA was using a 3D video motion capture system using two R6502 microprocessors.

Very interesting... just to record some points from the document:

- authors J H Nixon and J P Cater of SwRI (Southwest Research Institute, San Antonio, Texas)
- 28 May 1982

- third generation of such experimental systems in a series for NASA

- recording limb position (range of movement, dexterity) of a human subject in a spacesuit
- subject is decorated with LEDs controlled by an off the shelf system
- retrieving data from the off the shelf system ("Selspot") which controls up to 30 LEDs and records position with 4 cameras
- off the shelf system has a digital interface
- those cameras have a full-frame infrared detector and some analogue means of measuring the 2D position of the single spot on the detector, digitised as a pair of 10 bit numbers
- 300 samples per second, approx
- the positional error is about 1 inch, on the scale of a human subject

- recording and replaying binary data from videotape, at four bytes per used line, a 1 Mbit/sec rate
- variable number of lines used per frame
- remainder of frame is conventional video of the experimental subject
- data is packetised with a one byte checksum
- off the shelf (Panasonic VHS) VCR, solenoid controlled, digital interface to "remote control cable"

- written in 6502 assembly code, listings also submitted and possibly on file
- two R6502 subsystems, at 1MHz and 2MHz respectively, for central control and for VCR data transfer
- two 1k buffers, single port, flipping between the two systems for moving data between them
- AMD9511 for floating point support
- each data frame recorded with a frame counter, dropped frames reported as an error to the user
- machine code monitor on a serial line interface
- GPIB/IEEE 488 bus interface for data export, in first instance connected to an HP 9825 desktop
- user interface has a VDU output driven by a TMS 9918 chip with 16kB of DRAM

- some difficulty reconciling the "frame rate" of the position sensor system with the VCR's frame rate.
- initially hoped to store 3x to 4x the data on the VCR

Perhaps see also

I'm aware of a system developed in 1987 to perform quality control for glass-blowing. It performed a digital, one dimensional convolution on similar input. A 1982 system could approximate this with delay lines and operational amplifiers. Glass-blowing is the easier task because there should only be two sharp edges per monochrome, horizontal line. With motion capture, there may be two or more pairs of close, vague edges per line.

_________________
Modules | Processors | Boards | Boxes | Beep, Beep! I'm a sheep!