COMMENT ⊗   VALID 00006 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\input macro.tex[let,jmc]
C00011 00003	III A 2.
C00015 00004	IV A.
C00021 00005	V.
C00024 00006	VI B.
C00027 ENDMK
C⊗;
\input macro.tex[let,jmc]
\nopagenumber
III C

	Dr. Binford's current research includes machine perception, robotics,
geometric modeling and computer graphics.
Besides its fundamental scientific importance, this research has applications
to problems of biological perception, to
computer-aided design and manufacturing, work
in hazardous environments, rehabilitation, and cartography.

	His progress in determining edges in images contributes to a mathematical
understanding of the function of lateral inhibition and simple cells
of Hubel and Wiesel.  He is investigating the linking of outputs of 
these edge operators and studying the architecture of these operations
for implementation in VLSI.  These results have potential for major
improvements in computer vision equipment.

	He also investigates automated stereophotogrammetry, in which the
central problem is determining surface interpretations to sequences
of images with partial correspondence of photometric and geometric
properties.  There is a very large search space among ambiguous possible
interpretations.  He has derived strong geometric constraints for
general surfaces and for special surfaces.  Search procedures based
on dynamic programming are being investigated.  An integrated stereo
system is being designed to support further vision research.

	Dr. Binford studies geometric representation of object shape and
its use in interpretation of images.  Geometric cues from single images
for segmentation of objects in images and establishing depth relations
among surfaces are under investigation.  He is concerned with inferring three-space
shape from single images.  These topics relate to understanding
of Gestalt perceptual phenomena, in understanding interpretation of line
drawings, and in applications.  These capabilities are being integrated
in a system for geometric reasoning, ACRONYM.  Research is underway to
introduce a formalization of geometry in ACRONYM.

	He also supervises and participates in research on programming language
for robots and dynamics and force control.

	Our expectation that he will be successful in this research
is based on his record
of the last ten years.  During the past five years, his
group has been widely recognized as one of the world's leading
robotics research groups.




\vfill \eject
III A 2.

Statement by John McCarthy, Professor of Computer Science, on the
important papers of Thomas O. Binford.

	Progress in computer vision depends on the representation of visual
information in a computer.  This information must be represented in a way
that will facilitate recognizing a given object in a variety of
environments and determining the locations of its parts so that it can be
manipulated by robotic and other computer controlled devices.

	The easier vision schemes represent only two-dimensional objects or
polyhedral objects.  This is sufficient for some purposes, but fully
flexible vision systems must represent objects as three-dimensional
entities in such a way that they can be recognized when viewed from
arbitrary directions, with arbitrary placement of the parts if they are
articulated objects and when parts of the object are obscured by other
objects or other parts of the same object.

	Binford's 1971 paper and the papers with Agin (1973), Nevatia (1977), and
with Brooks and Greiner (1978) represent continued progress in developing
Binford's representation of three dimensional objects by systems of
generalized cylinders and cones and the use of this and related
representations by recognition and manipulation programs.  The generalized
cone representation is the most ambitious in current use for representing
three-dimensional objects.  The letters from Kanade, Horn, Winston,
Shirai, Brady, Bajcsy and O'Callaghan all point out the importance of this
work.

	Binford's 1979 "Computer Integrated Assembly Systems" presents
a plan for improving the capabilities of assembly robots to include
touch and force sensors and to give the programmer of the robot
a programming language that makes feasible programs that use these
capabilities and can co-ordinate two arms and co-ordinate manipulation
with vision.  This work is a key contribution to increasing industrial
productivity by opening new applications to robots.

\vfill \eject
IV A.

	Dr. Binford's role as Adjunct Professor will have two aspects.  First
he will continue his personal research and direction of students.
Second, he is expected to play a leading role in maintaining Stanford's
position in research in industrial robotics as applied to the problem
of productivity.  This problem is attracting increasing attention in
industry and government as other countries, especially Japan, have
passed the U.S. in industrial productivity in many fields.  The Japanese
success has involved much more extensive use of industrial robots
than in the U.S., and our efforts to increase productivity will
also involve robotics.  The Stanford Artificial Intelligence Laboratory,
whose robotics work has long been directed by Binford, was one of
the initiators of sophisticated robotics work in the world and has
maintained a leading position.

	Professor Cannon says in his letter of recommendation,
{\it "I am hopeful that we will be
mounting a strong Stanford program in productivity.  Tom would
certainly play an important part in it"}.
While an organizational form for an enlarged Stanford program in
productivity has not yet been proposed, there has been wide agreement
on the need and opportunity for such a program among faculty in
the Computer Science, Mechanical Engineering, Electrical Engineering and
Aeronautics and Astronautics.  The computer science faculty is
also of the opinion that Dr. Binford is essential for this program.

B.

	Binford has a proven record of research accomplishment as
shown by the attached letters, he has supervised many students who
are among the best in the field, and he has had no trouble in finding
100 percent funding for his work.

C.
	Binford's work in robotics has involved substantial interaction
with the mechanical engineering department in the design and construction
of robotic arms and in the programming of their control.  The Stanford
arms served as prototypes for the Unimate 2500 arm which has recently
begun commercial use.

D.

	The plans for increasing Stanford's research in productivity
will involve extnsive interaction bbetween Binford's group and other
departments.  The letters from Professor Bernard Roth of Mechanical
Engineering and Robert Cannon of Aeronautics and Astronautics are
relevant to this.

\vfill \eject
V.

	While Binford's role has not been primarily as a teacher, he as
taught and will continue to teach a course in robotics and computer
vision, Computer Science 227.  This course attracts about 15 students each
time it is taught, and has received increasingly favorable ratings as
Binford's teaching experience has increased.
Student ratings of the course are attached.

	Many recent PhD advisees of Professor John McCarthy in vision and
robotics received a major part of their guidance from Binford.  It is
expected that he will take over this role almost completely for students
in vision and entirely for students in robotics.

	The following scientists whose dissertations were done in
Binford's group and with substantial guidance from Binford are leaders in
academic and industrial research.

	Dr. Russell Taylor (1976), IBM, has played a large part in
development of programming systems for robots at Yorktown Heights and in
their development of a robot product for internal use at Boca Raton, Fla.

	Prof. Ramakant Nevatia (1974), associate professor with tenure,
Computer Science Department, USC, shares leadership of the Image
Understanding research program.

	Prof. Ruzena Bajcsy (1972), professor of Computer Science,
University of Pennsylvania, leads a large group in biomedical image
analysis, computer vision, and robotics.

	Prof. Gerald Agin (1972), biomechanics department, Carnegie Mellon
University, has led the industrial vision program at SRI, developed the
software for the SRI vision module, and plays an important role in the
Robotics program at CMU.

	Dr. Robert Bolles (1976) leads the program in industrial vision at
SRI.

\vfill\eject
VI B.

	At the request of the Computer Science Department, the
evaluations of Binford's work were obtained by Professor John
McCarthy.  Apart from that, there wasn't a formal committee.
Enclosed is a copy of the letter sent, a list of the
addressees and the replies received.  Since Binford's work is
internationally known, it was possible to obtain evaluations from
the leaders of all the major centers of research in AI and robotics.

	Replies were not received from Bernard Chern, Jerome Feldman, and
George Giralt.

	The tenured faculty of the Computer Science Department read
all the letters of evaluation and voted unanimously to recommend
the promotion.  This includes the two members who were absent from
the meeting itself.

\vfill\end