1.8 SOME WELL KNOWN PLANNING SYSTEMS
Most CAPP systems have been through many revisions which can be illustrated through the simple grouping shown in Figure 1.22 Genealogy of Some Families of CAPP Systems. The order and relationships of these systems were found or deduced through the review of the literature discussed throughout this chapter.
The list below summarizes some systems for Process Planning. Longer descriptions are given for some of the more well known systems. Systems without names are referred to by the authors name. The summary of each system should provide an overview to many of the planning approaches taken to date.
APPAS, CADCAM [Chang et. al., 1982], TIPS [Chang and Wysk, 1985], QTC [Kanumurz et. al., 1988] [Anderson and Chang, 1989] [Chang, 1991] - APPAS (Automated Process Planning and Selection) models machined surfaces with special codes. Decision trees are used to direct the planner. It can handle multiple passes and operations for each surface. APPAS was extended into the CADCAM system to allow the use of an interactive graphic interface for design entry for direct use in process planning. This example was taken even farther in TIPPS that added a full graphical design system to allow the user to guide the CAPP system through a point-and-click interface. And, subsequent development lead to QTC (Quick Turnaround Cell). This system is based on a CSG type recognition system that plans from design to production.
AUTAP, AUTAP-NC [Eversheim et. al., 1980] - AUTAP will first find a stock shape, then develop a set of operations to produce rotational parts. Operations are sequenced, time estimates are made and tools are selected. AUTAP-NC is a subsequent system capable of producing NC part programs to produce the turned components.
CADCAPP [Kamvar and Melkanoff, 1986] - Describes a system that accepts a drawing of a turned part from a commercial CAD system, then uses an algorithm to assign a GT code.
CAPES [Fujita and Oochi, 1989] - A generative system for prismatic parts. Uses a solid modeler.
CIMS [Iwata and Sugimura, 1985] - Uses interactive solids modeling with associated technological information. A process plan is generated using the finished part, and a blank. The plan is generated using production rules in a knowledge base. The planning is done in a five step process: machined surfaces are identified, appropriate tools are selected, machined surfaces are grouped, precedence constraints are generated, and finally, a process plan is generated. Sequencing is done using matrix methods.
CMPP [Dunn, in Nolan ] [Parks et. al., 1989], XPS-I [Pavey et. al, 1986] [Groppetti, 1986] [Austin, 1986], XPS-II - For rotational, and other domains such as grinding, based upon representation. The planner analyses the machinability, then produces plans. This system contains the language COPPL described earlier in the chapter for manufacturing knowledge. A variety of research has been done on applying this system, such as accepting IGES files as input to eventually generate NC code [Knutilla and Park, 1990].
CUTTECH [Barkocy and Zdeblick, 1984] - Uses information from the user about features, machines, materials, etc. Generates a GT code for each feature, selects machine tools, cutting tools, etc. using rules from a knowledge base. A second knowledge base is used for selecting operation parameters like speeds and feeds.
[Delbressine, 1989] - A system to develop a part using stock that is transformed using manufacturing objects based on swept tool geometries.
EXCAP [Davies and Darbyshire, 1984] [Joseph and Davies, 1990] [Davies et. al., 1985] [Darbyshire and Davies, 1984] - Uses a limited description for rotational parts. Planning is divided into two steps, sequence planning, then operation planning. Rules are backward chained to prove hypothesis. If accepted the geometry is modified, and planning continues. This stage constructs a tree where the root is the finished part, and the leaves are blanks. The final tree has branches with certainties associated with each. Plans can be easily generated from the tree, and if rejected, alternate plans are easily generated. A further exploration with an application was done by Joseph .
EXPLAIN [Prabhu et. al., 1990] - A feature based system that generates process plans for turning using rules.
EXPLAN [Warnecke et. al., 1989] - Describes a generative planner using a graphical interface.
FEXCAPP [Lee et. al., 1989] - Uses connectivity graphs and then does pattern matching to find features.
FLEXPLAN [Lampkemeyer et. al, 1991] [Tonshoff et. al., 1989] - Uses Petri nets to represent non-linear process plans. FLEXPLAN can store, generate, and schedule for non-linear process plans.
FRAPP [Henderson and Chang, 1988] - A system that does B-Rep based feature recognition, then some process planning based on the features.
GAPP [Laperriere and ElMaraghy, 1992] - An assembly planner using part connectivity and an A* search.
GARI [Descotte and Latombe, 1984] - Parts are described with feature descriptions, dimensional locations and tolerances of features, and other information like materials and finishes. An initial loosely constrained plan is developed, then a hierarchical planning system is applied in an iterative manner to suggest operations, propagate constraints, and backtrack.
GCAPPS [Pande and Palsule, 1988] - A system for feature based design of rotational parts.
Genoa [Held and Juttner, 1991] - Uses rules to generate plans in a limited domain.
[Han et. al., 1987] - A semi-generative system.
[Henderson, 1986] - A method for recognizing holes, slots and pockets. Faces are identified, then their relationships are determined, and converted to solid features with a connectivity graph.
Hi-Mapp [Berenji et. al., 1986] - Parts are represented with feature descriptions, locations, tolerances, and other manufacturing information. An abstract, but correct plan is generated. A hierarchical planning system is applied, which operates using priorities. The knowledge base contains rules for selecting processes, recommending cuts and machines, and recommending other user defined options.
KAPPS [Iwata, 1988] - An expert process planner for machining using over 600 rules.
KCAPPS [Wei et. al., 1990] - An AI generative system with an interactive interface for defining manufacturing attributes. Uses CSG subtractions of features from various sources of stock.
[Lee and Wang, 1990] - Does robotic assembly planning. They use features and rules to get a) largest feature stability, and b) heaviest part, and then rank the assembly order. It then generates robot path parameters for a complete off-line programming system.
MicroCAPP/GEPPCAPP [Wang and Wysk, 1985 a b] [Wang, 1984] [Cachia and Vajpayee, 1986] - Complementary systems that have MicroCAPP, a code based variant planning system, and GEPPCAPP, a decision tree based generative planner.
MiniCIM [Shyu et. al., 1987] - A complete design to manufacture system for turning.
PART [vanHouten et. al., 1984] [Jonkers and Kals, 1992], PART-S [deVin et. al., 1992] - A generative system that does both process and production planning. Another implementation is PART-S for doing sheet metal planning.
PC-CAPP [Pande and Walvekar, 1989] - A feature based CAPP system for prismatic parts for a specific corporation. Primarily a rule based data-base lookup package.
[Phillips et. al., 1985] - Discusses a profile based planner.
[Preiss and Kaplansky, 1984] - A system that goes directly from design to NC milling.
PROPLAN [Phillips, 1984] [Phillips and Mouleeswaran, 1985] - Primitive lines and arcs, describes a symmetrical rotational part. Production rules are stored in a database and applied using a graph search. The graph nodes are the parts geometry, the arcs are the transformation operations.
PWA_Planner [Chang and Terwilliger, 1987] - Does assembly planning for circuit boards.
ROUND [vanHouten and Kals, 1984] - A generative system for turned parts.
SAPT [Milacic, 1985] [Milacic and Vrosevic, 1984] [Milacic et. al., 1988] - Three modules are used, the first is technological pattern recognition, which defines detailed structure of process planning. Secondly, the manufacturing process logic module defines logical relationships between the part, machine, fixture, and operation sequence. The rules for these tasks are stored in a knowledge-base, along with rules for the strategy of planning.
[Schneider et. al., 1989] - Their system attempts to match rules for rotational parts by examining the 2D profiles.
SIPP [Nau and Chang, 1985] - Hierarchical frames are used to represent parts, where a frame will represent a feature, and other frames define the properties of the feature. A search strategy using least cost first, is used in conjunction with a knowledge base relating features, to machining processes. The nodes of the graph represent parts, and the arcs are the operations that transforms them. Process restrictions are considered in generation of the graph.
TOM [Matsushima et. al, 1982] [Iwata and Sugimura, 1984] - Technostructure Of Machining uses a final design, a collection of holes, and backward chains a set of production rules to develop an optimal process plan. When conflicting rules are found heuristics are used, including frequency of application.
TOPS [Pinte, 1987] - A system that uses frames, constraints and rules with a truth maintenance system for planning.
[Willis et. al., 1989] - A system that uses a solid modeler based on an expert system. Non-linear planning is supported.
XCUT [Hummel, 1989] [Hummel and Brooks, 1988] - Uses a hierarchical layering of rules Even more important is the use of Meta-knowledge to select between modules of rules.
XMAPP [Inui, 1986] - Does forward planning using a product model like GARI.
XPLAN [Alting, Zhang and Lenau, 1988], XPLAN-R [Zhang and Alting, 1988 a, b] - A planning system for rotational parts based on DCLASS. The successor is XPLAN-R.
XPS-1 [Groppetti and Semeraro, 1986], XPS-2 [Nolan, 1989], XPS-E [Chryssolouris et. al., 1986] - A interesting line of generative CAPP systems have been developed from the CMPP project, eventually evolving into the XPS-E system. CMPP (Computer-Managed Process Planning) was originally developed at United Technologies [Mark Dunn in Nolan, 1989]. It is a process planning system which is Generative, and Automatic, and capable of handling rotational parts. After the data entry module is done the Process Planning module will begin. Its four functions are; generate a summary of operations, select tolerancing datums, determine dimensions and tolerances, output process documentation. The Computer Aided Manufacturing-International (CAM-I) group has been developing a process planning system. This has been done through a set of projects dubbed XPS-N [Sack authored a section in Nolan, 1989]. The first system in this line was XPS-1, which was completed in 1984. The system provides an environment for process planning, using feature based product data, and manufacturing resource data. This information is stored in a relational database. Rules are used to provide the logic for process planning. The second prototype system, XPS-2 was completed in 1987. Other CAM-I design projects were used as an interactive front end for collecting feature information including, geometry, dimensioning, and tolerancing. An attempt to use Artificial Intelligence was proposed with XPS-E. This was done using a report commissioned from two French organizations, Institut National Polytechnique de Grenoble (INPG) and Industrie et Technologie de la Machine Intelligente (ITMI). Their method conformed to the ideal of the XPS-N Philosophy, but augmented it with Artificial Intelligence techniques.