ABSTRACT 'zhis paper presents an automatic assembly planning system which takw
the CAD description of a product as input and automatidy gene" an
assembly plan subject to the resoma constraint of a given assembly cell.
The system improves both the flexibility and productivi@ of flexible
manufacturing systems. and is composed of five modules: world datab.
simulated world model. knowledge acquisition mechanism, p-g
knowledge base, and assembly planner. The world database contluns the
information from a designer which provides the information about the p
duct and the assembly cell. The knowledge quisition mechanism acqb
knowledge neccssBty for assembly planning that is, the pnecedence
knowledge among assembly tasks, the fix- specification for the assem-
bly, and the tool requirement for the assembly tasks. The acqllired
knowledge forms mC planning knowledge base. The simulated world
model keeps lrack of the current state of the assembly world. In the initial
state, all the components are xpamted, while in the final state, all the com-
ponents are assembled together. The assembly planner is made up of a set
of production rules which models the effects of rcal assembly tasks. By
repeatedly applying these poduction rules to the simulated world ~Eate, the
pbna transfams the initial state into the final state. The set of rules
applied during this transformation pmcess forms the assembly plan to actu-
ally assemble the product in the given Bssembly cell. Examples are given to
illustrate the concepts in these five modules.8427
1. Introduction
Due to the inherent complexity in the assembly process planning. it is
usually carried out in a hienrrchical fashion [la] for robotic assembly ceU
There are four planning levels in this hierarchy: assembly-level planning.
task-level planning, manipulator-level planning, and pint-level planning.
Extensive work has been done from the joint-level planning up to the task-
level planning [2,7, 12,131, but relatively few literatures have talked about
the automatic generation of assembly plans. Bourjault [31 proposed an
algorithmic method of generating all the assembly tank sequences from a set
of precedence relationships among assembly mating tasks. Here an assem-
bly mating task refers to an operation or proms of putting two subassem-
blies toghr. The precedence relationship are obtained by interactively
answering two sets of st"d questions about the mating tasks. The
number of these questions is exponential with respect to the number of mat-
ing tasks. De Fazio and Whimey [5l improved Boujault's method by
reducing the number of questions to twice the number of mating tasks.
Wolter [16] proposed a so-called "constraint graph" to implicitly represent
the precedence relationships among the movements of components in an
assembly. Feasible assembly sequences can be generated by solving this 自动装配规划系统英文文献和中文翻译:http://www.751com.cn/fanyi/lunwen_6791.html