Keith A. Pray - Professional and Academic Site
About Me
·
·
·
LinkedIn Profile Facebook Profile GoodReads Profile
Professional
Academic
Teaching
                                          
Printer Friendly Version

Intro ] [ 01-DPMED ] [ 02-Dominic ] [ 03-DSPL Air-Cyl ]

Up: AI in Design ]

Critique: Dominic
Dixon, Howe, Cohen, Simmons *

      The paper flowed very nicely, starting with a few words about what design actually is and what processes can be / had been described. This was followed by defining the terms they intended to use. It made things easy to follow, although the dependency table was missing from the problem formulation list. Since it couldn't have been my fault, the initial confusion I experienced from the dependency list being static must have been due to the dependency table not being introduced in the same list.

      In the description of Dominic's architecture there is an inference engine. I found it odd that there was no mention of rules. Granted the dependencies, dependency list and table, performance satisfaction scales and constraint equations can be viewed as a form of rules. The authors identify two other systems as performing better than Dominic. This is most likely due to Dominic's lack of a way to represent domain specific knowledge relevant to the design approach. DPMED seems to remedy this slightly, although its main concern is not with domain independence.

      Despite being in the middle, chronologically, of this week's group of readings, this paper serves very well as an introduction to the concept of redesign. This seems to be due to Dominic resembling an optimization tool more than a design tool. This restricted problem space leaves room to define the concept of design goals (problem specification parameters), output of the design tool (design variables), analysis (performance parameters), assessment (performance satisfaction scales), etc.

      Many of the problems of Dominic manifest themselves in DPMED, including those related to the hill-climbing approach.

      It seems very inflexible to have the best value on a performance satisfaction scale be the mean value of the allowed range. An approach allowing equations representing a scale or sections thereof might overcome this limitation.

      Overall, I wasn't impressed with Dominic's progress in achieving domain independence. The domains explored seemed to be all slight specializations on the general mechanical design domain. It could be that I studied DPMED in depth before Dominic. DPMED came about two years later and with the knowledge gained from work done with Dominic I & II. This paper does present the role of iteration in the design process and basic concepts well.


* J. R. Dixon, A. Howe, P. R. Cohen, M. K. Simmons, "Dominic I: Progress Towards Domain Independence in Design by Iterative Redesign," Proceedings, 1986 ASME Computers in Engineering Conference, Chicago, IL, July 20-24, 1986 (with E.C. Libardi and M.K. Simmons).

Intro
01-DPMED
02-Dominic
03-DSPL Air-Cyl
04-Pride
05-COSSACK
06-MICOM-M1
07-Configuration Survey
08-Dynamic CSP
09-MOLGEN
10-Failure Handling
11-VT
12-Conflict Resolution
13-Cooperative Negotiation
14-Negotiated Search
15-Multiagent Design
16-Prototypes
17-CBR Survey
18-PROMPT
19-A Design
20-Bogart
21-Cadet
22-Argo
23-Analogy Creativity Survey
24-Algorithm Design
25-AM
26-Edison
27-LEAP
28-Plan Compilation
29-ML Survey
30-Strain Gauge
31-Grammar
32-Config GA
33-Functional First
34-Functional CBR
35-Functional Survey
36-Models
37-First Principles
38-Config Spaces
39-Task Analysis

by: Keith A. Pray
Last Modified: August 13, 2004 11:47 PM
© 2004 - 1975 Keith A. Pray.
All rights reserved.

Current Theme: 

Kapowee Hosted | Kapow Generated in 0.009 second | XHTML | CSS