00100			A DISSERTATION PROPOSAL
00200	
00300			by:  Vic Scheinman
00400			December 6,1973
00500	
00600	
00700			THE COMPLETELY AUTOMATED MACHINE SHOP
00800	
00900		The  traditional  corporate  setup  for  advanced  mechanical
01000	design  and  prototype  engineering  groups is Engineer-Designer-Shop
01100	foreman- Machinist.  This structure usually means that the turnaround
01200	time  from  idea  to finished prototype is much longer than ideal for
01300	most creative and active design engineers.
01400	
01500		By adding a computer into the loop, and training the engineer
01600	to  use it properly, great economies both in manpower and in time can
01700	result. In the simplist form, the computer acts as a bookkeeping aid.
01800	On  a higher level, the computer serves as a designer by enabling the
01900	engineer to put ideas on paper in drawing form suitable  for  machine
02000	shop  use.    In the next higher form, the computer can take over the
02100	shop foreman's task of selecting  or  ordering  the  proper  material
02200	stock   and  planning  the  required  machine  operations  and  their
02300	sequence, including time estimates and  optimizing  the  use  of  the
02400	available  equipment.  In the almost ultimate form, the computer can,
02500	with the  aid  of  suitable  computer  controlled  hardware  devices,
02600	actually  perform  the  required  machining operations and subsequent
02700	inspections necessary to completely produce the designed part. It  is
02800	towards this end that this proposal is directed.
02900	
03000		Of  course,  this  problem has been tackled before, and it is
03100	even being worked on at the present time. Various  organizations  and
03200	governments  are  presently putting large sums of money into research
03300	on complete automation.  This work is now to a  point  where  I  feel
03400	that  a  working  and  reasonably  versatile version of the automated
03500	machine shop is a reality, and some form of it is within the scope of
03600	this proposed dissertation.
03700	
03800		Lets  now look at the problem and get an idea of where things
03900	stand at the present time and why this topic is a realistic  sort  of
04000	thing to be working on for a PhD.
04100	
04200		Since  designing  a  computer  controlled  electric arm as an
04300	Engineers Degree Thesis in 1969, I  have  produced  several  improved
04400	versions  of  this  device.  Concurrently, other graduate students in
04500	the Computer  Science  Department  have  been  working  on  operating
04600	programs  for  the  arm.    At  the  present  time,  it is capable of
04700	performing numerous simple assembly tacks following  a  preprogrammed
04800	outline, but accomodating the real situation based on perception thru
04900	force touch and visual feedback.  A  programming  language  (HAL)  is
05000	presently  being  developed  to  give  arm  programing ability to the
05100	computer oriented engineer.
05200		At the Stanford A.I. Project, our display system is also at a
05300	relatively advanced state, when compared with other  universities  or
05400	industrial  organizations.  As a product of several recent theses, it
05500	is practical to generate  and  manipulate  geometric  shapes  on  the
05600	display  system  in  addition to being able to handle traditional 2-d
05700	display representations.  The process of creating and handling  these
05800	shapes  has  also  been  reduced  to  a  special  purpose programming
05900	language. It is thus evident that a  number  of  the  basic  building
06000	blocks  for  this  proposed complete system already exist. Certainly,
06100	one of the immediate  problems  is  that  of  developing  a  suitable
06200	planing   and  supervisory  program  which  will  coorinate  all  the
06300	subprograms and perform all the proper calls required to get the idea
06400	into  the  shop. Then, a good machine tool programming method must be
06500	developed.  No doubt, this should  draw  heavily  from  the  industry
06600	accepted  standard  N-C Programming language APT.  Continuing on with
06700	the system, there is the execution part of the program;  the  machine
06800	tools.    It  is  not clear that just getting an N-C mill or lathe or
06900	punch  will  be  enough.    Setting  up  the  part   for   machining,
07000	transferring  the part between machines, or even reorienting the part
07100	all appear to be tasks which are done by hand, even on  most  of  the
07200	current large N-C machines.  The use of a mechanical arm as a general
07300	manipulating device to perform  many  of  the  human  type  functions
07400	appears  to  be  a  good  solution.   In concept form, the shop would
07500	consist of a number of  N-C  type  or  remotely  controlled  type  of
07600	machine  tools all comtrolled by a computereither directly or through
07700	a mechanical arm which would perform the  necessary  setup,  control,
07800	and inspection tasks, in lieu of the human operator.
07900	
08000		To  be fair, I have to confess to having a particular type of
08100	machine shop in mind at the present  time,  as  the  concept  of  the
08200	totally  generalized  machine  shop  is  just  too great a problem to
08300	tackle at the present time.
08400		The  demonstration  shop  I  have  in  mind would be oriented
08500	towards handling sheet metal type of parts.  It would be  capable  of
08600	cutting,  bending, punching and spot welding sheets of aluminum.  The
08700	shop would contain ,in the initial concept, the following tools.    A
08800	shear for cutting sheets of aluminum, a brake for bending aluminum, a
08900	punch or exuivalent such as an N-C mill for punching or cutting holes
09000	in  the  material,  a  spot  welder  for fastening pieces of material
09100	together, and several special purpose tools such as a notcher , and a
09200	versatile  clamp. It should be mentioned that the sheet metal shop is
09300	one area where few machines have been fully automated at the  present
09400	time.   As the universal manipulating and handling device I propose a
09500	semi-mobile version of the electric manipulator currently uun use  at
09600	the  A.I.   Lab.  A special set of fingers would probably be required
09700	too. As a typical initial goal, the making of randomly sized  chassis
09800	boxes  would be produced by this facility.  Simpler shapes could also
09900	be produced, but shapes involving multiple separate pieces  could  be
10000	excluded.   Thus, in concept form, anyone could sit down at a display
10100	console and design a chassis, bracket, simple  sheet  shape  or  form
10200	with  a  control program operating interactively to give the designer
10300	feedback as to whether the design is possible, how complex it is, how
10400	many  different  operations  it will take, and how long it will take.
10500	Also , some displays of the blank  shape,  allowances  for  handling,
10600	bending stretch, punch minimum radii, etc.