Friday, 23 June 2006

Robot Parts (4/4)


A series of product design for robot assembly seminars were held at a UK university.  They were well attended and the object of these seminars was to encourage industrialists to analyse their products, using a product design for robot assembly computer software application.  The results of these studies were then investigated by university staff so that handling, gripping and insertion requirements for robot assembly could be recommended. These seminars were funded by the ACME directorate as a means of forging closer links between universities and industry. The results of the studies also gave direction to future research work at the university in the field of robot assembly.  Interested parties were given a copy of a computer software application.  Industrial product data was stored in standard ASCII files and this was easily manipulated by staff at the university.  Statistics were produced that indicated trends in parts geometry and the physical properties of parts.  These statistics showed the relative importance of various pieces of assembly automation for a cross-section of industrial products and they gave indicators for future assembly hardware development.


I strongly believe that industry will only demand products and services if there is a genuine need for either. For this reason, my direction is heavily influenced by the continuously changing needs of my clients. Product information (available to clients) and calculated information (demanded by clients) is monitored through my consultancy contracts.  My approach to the presentation of parts for robot assembly is changed to best suit the needs of the majority of my current clients and future clients. The results of the previously mentioned data acquisition seminars influenced the range of handling devices included in the database. It was necessary to include other devices to cater for particular categories of parts, that were thought to exist in smaller numbers than in reality. The findings also affected the handling expert system format. The sequence of questions was altered so that the minimum amount of information was required for the majority of parts.  Later, a consortium of six companies was being formed to interface the product design for robot assembly software with a conventional CAD system.  The object of this work was to allow a product designer,using a CAD system, to have the benefit of product design for assembly running in the background, which only became active when adverse robotic assembly properties were evident.


The presentation of parts is a topic often neglected by those considering robot assembly and yet it accounts for the majority of the cost for an installation. It is important to be able to describe the features of a part by the use of a parts classification technique that is sufficiently comprehensive to fully describe the part, without involving undue effort, or understanding, from the user.  Parts presentation devices for robot assembly should have a high general-purpose content and a low special-purpose content.  The orientation of the part during insertion affects the choice of handling device and the number of robot degrees of freedom. The classification of a part for handling can be a tedious process and it is important to only define features that are relevant for the selection of handling devices. This is best achieved by using an expert system approach and decision trees.  The complex process of handling device selection can be carried out by computer software applications, thus eliminating the need to manually carry out many iterative calculations. The types of handling devices which best suit the needs of industry can be chosen by asking current and potential industrial users to specify their particular handling requirements.  Most of the information relating to the design features of products, for robot assembly, can be extracted from a CAD system database.

No comments: