A Petri Net Framework For Representing Mechanical Assembly Sequences

Abstract

~Choosing a representation of assembly sequences that is flexible and can be used to compute a range of information is important when designing an intelligent control system for the assembly process. In this paper we propose an extended Petri Net formalism which can be used to model Mechanical Assembly sequences. The proposed framework can be used. to compute a range of information relevant to the assembly process by specifying the interpretation for the net. I. mODUCTION Robotic assembly systems have the flexibility to handle wide ranges of styles and products, to assemble the same product in different ways, and to recover from errors. An assembly plan for a given product consists of a set of assembly tasks with ordering constraints among its elements. Each assembly task consists of joining a set of sub-assemblies to yield a larger sub-assembly. A sub-assembly is a non-empty set of parts that either has only one part, or is such that every part has at least one surface contact with another part in the subset. Such contact between parts are known as connections [l]. A connection can be in an unconnected or a connected state. Given an assembly plan, an assembly sequence is an ordered sequence of the assembly tasks that satisfies all the ordering constraints. The assembly process starts with separated parts that are joined together based on the connections specified to form.the complete assembly. The sequence of assembly tasks and the cost of an assembly task (in terms of time consumed for the task and complexity of the assembly task for example) are some of the types of knowledge pertinent to assembly plans. An intelligent robot malung decisions about assembly plans has to compute these kmds of information from the model or representation of the assembly it is given. Other kinds of knowledge including the state of the connections between two sub-assemblies, the sequence of assembly tasks involved may also have to be computed by the robot. Some of these knowledge is dynamic in nature in the sense that the knowledge gets updated at each step of the assem