**************************************************************************
**************************************************************************
**
**
** Project by   :  Lin Chia-Chen
**
** EMAIL        :  c473bf
**
** Date	        :  May 28, 1991
**
** Domain name 	:  autoflight  (automatic pilot)
** 
** Solvable    	: Given the initial state of an aircraft, the destination,
**		  the airport traffic, the auto pilot should fly the 
**		  plane to the destination safely and be able to handle
**		  weather conditions.
**
** Hours       	: --
**
** Reference   	: a. Prodigy Manual
**		  b. Incrementor for the rudder is obtained from the mail
**	   	  message by D. T. Orman
**
** Status     	: This planner (the auto pilot) finds the flight path from
**  one city to another, checks the weather conditions through the course of 
**  this path and sets all necessary instruments to handle these weather
**  conditions.  It tries to find the flight path within a country's
**  boundary first and if no flight path of this sort is present it will 
**  then cross the boundary lines.
**
** Modelling difficulties:
**  
** (A) 	a. The model of the automatic flight pilot presented in this
**	   domain does not model the real world aspect of *TIME* dependency.
**         Due to the lack of methods or means to simulate what switch or
**	   control should be turned on at a particular time.
**
**	b. The trajectory of the plane is three dimensional.  This model
**         disregards totally the difficult task of actually simulating
**         this in a symbolic system. (it's hard enough to write an integer
**	   incrementor in Prodigy)
**
**	c. Precision of the controls such as the stick due to its continum
**	   aspect were modeled discretely.  (eg: how would you represent
**   	   the position of the stick at the position (x, y) = (0.3, 0.8) )
**         Coherence of some other devices such as the position of the
**     	   flipper and the rudder are dependent on the position of the 
** 	   stick and other instruments in the real plane.  And this again
**	   raises another difficulty for this domain.  (Note: the turning
**	   of the rudder in increments of one is a crude attempt to 
**	   simulate this continum)
**
** 	d. Weather conditions in this domain are predetermined with 
**  	   100% accuracy in this domain, which is not at all what modern
**  	   technology can achieve.  Human pilots must be experienced in
**  	   all sorts of weather condition and flight hours to obtain the
**  	   skill necesary to handle unexpected weather conditions.
**	   Unless we could develop a system better or equal to the human 
**	   input and output system, automatic pilot that could be confided
**	   in flying an aircraft would be inconceivable.
**  
**  	e. Finding the shortest path is not made an issue in this domain.
**	   Although repeated cycles are handled by control rules in this
**	   domain, shortest path are not guaranteed.
**
** (B)     Before writing control rules, I didn't know that some of 
**	   the problems are unsolvable in 2500 nodes.   But after
**	   writing the control rulers I discovered that control rulers
**	   did more than what I expected.  (it avoided the cycles
**	   in my graph, even though all I originally planed, was to make
** 	   the plane take off and land on appropriate cities.  See
**	   the comments for the PREFER-FLYING-IN-THE-COUNTRY-OF-DESTINATION-CITY
**	   search control rule).
**
** (C)	   The interface to the UNIX file system were particularly 
**	   easy.  Domains with their problems could be developed entirely
**	   independently of each other and the access of these are
**	   also made easy by the domain and problem commands at the 
**	   prodigy interpreter.
**
** (D)	   Prodigy's analyze mode could be improved for those people
**         who do not have access to X window system.  Often times
**	   the information returned to the user is too scare to allow
**	   efficient debugging.  Regardless of the fact that some 
**	   information about the node could be retrieved if the user
**	   tries hard enough to learn all the commands. (eg: at the failure
**	   node, prodigy only returns with a failure node message which
**	   is useless for debugging purposes.	
**
**	   The other problem is the increadibly small amount of feedback
**	   returned when writing control rulers.  Basically, there is 
**	   no help at all from the system for debugging control rulers.
**
**	   Another thing that could be improved is the TEXT mode.
**	   Make be the output of the trace could be dumped into a file
**	   instead of to the screen.  This would allow the user to look
**	   at the trace more easily and at a time that is more convenient.
**
***************************************************************************
***************************************************************************


***************************************************************************
***************************************************************************
**
**  Note: The outputx files in this directory are the output with/without
**  	  search control rules produced by this domain.
**
**  Note: To see a list of functional problems please refer to the
** 	  README file in the  ./probs subdirectory
**
***************************************************************************
***************************************************************************



