Cart Vision - November 1971 to November 1972 to November 1973.

                           B. g. Baumgart

	My cart Vision work consists of a  Geometric  World  Modeling
program,  GEOMED;  and a television picture anaylsis program, CAREYE.
Many essential smaller programs, such as camera  locus  solving,  are
embedded  in  these  two  large  programs.   My  approach  to  visual
perception may be characterized as "extreme  mechanical  deteminism",
and is based on three assumptions:

i.  I assume that given the location of the TV camera and the sun and
given a faceted (not necessarily  planar)  polyhedron  model  of  the
world  a  computer  can  be  made  to compute a television image very
nearly like the one produced by a real digital TV camera looking at a
real world under a real sun.

ii.   I  assume  that  given  two similar TV pictures (or region-edge
pictures) a computer can be made to find the substantial similarities
and  the  slight  differences.   And that the slight differences in a
single image can be used to correct the camera and world models,  and
that  slight  differences  accumulated  over  many  images  will  add
substantially to the world model and indeed may be  characterized  as
visual perception.

iii. I assume that the `tabula rasa', world model initialization, can
be  done  with human assisted programs. The process of constructing a
computer model of the mundane physics and geometry of the real  world
is  is  not  fully  understood;  and  inspite  of recent successes in
automatic strategy planning it is  difficult  to  automate  a  poorly
understood process.

	Over the past year CAREYE has been recoded twice  and  GEOMED
has  been recoded four times.  In GEOMED a new hidden line eliminator
and geometric data structure have been developed;  while  in  CAREYE,
the  design strategy which was to convert TV image data directly into
region-edge data has evolved a middle step of  converting  TV  images
first  into  video intensity contours, similiar to those of krakauer,
and then into region-edge data structures. Also this year I wrote  an
alternate  reference  manual  for  Micro Planner and assisted Richard
Orban in installing my  hidden  line  eliminator,  OCCULT,  into  the
Hand-Eye system. The present cart vehicle is in working order and has
been exercised from time to time under manual remote radio control.

	During  the  next  year,  I  anticipate  no  new ideas and no
recoding. I will implement my present ideas  within  existing  system
designs. In the coming year I will get my hidden line eliminator that
predicts shadow edges and regions to work, and I  will  get  my  body
locus  solver  that  constructs  3D polyhedron models out of 2D image
regions from several cameras to work, and I will get  my  two  camera
locus  solvers  to work (one based on region matching the other based
on landmark-feature point matching).