Annual Report Department

2.5CAD/CAM, process and production planning

cad, cam and process planning

2.5.1Tool path planning and generation for multi-axis machining of complex shaped products

Multi-axis milling can improve drastically the machining process for complex shaped surfaces. More multi-axis machinery is being installed in the metal working workshops nowadays. Despite this, commercial CAM systems to program those machines still lack a lot of support functionality to plan multi-axis operations fast en efficiently.

This research aims to develop advanced strategies for the multi-axis machining of complex shaped parts. In the early nineties, the research was concentrated on tool path generation. Algorithms for dynamic optimisation of the milling cutter orientation with respect to the complex shaped surface were developed and implemented. These algorithms proved to be very successful and were further integrated into a CAD/CAM system. The functionality was enhanced with optimised finishing operation planning and collision avoidance strategies.

Because the STL data format for product geometry description is gaining importance and seems to be quite useful for representing complex shaped parts, the optimal tool path generation algorithms were extended to work on STL files as well. Last year achievements in this respect was the proof testing of the software developed for optimal gauge-free tool path generation for STL.

Besides tool path generation, the scope is broadened to full operation planning for the multi-axis machining of complex shaped parts. The objective is to automate the selection of operations needed to machine the part, i.e. from roughing till finishing. Last years progress in this area was the development of segmentation strategies to segment the complex shaped part geometry into complex machining features. Also a system for setup evaluation and selection for the complex machining features has been deployed. Qualification mechanisms in order to calculate the appropriateness of several possible finishing operation types for particular features are developed. All these functionality is supportive for the multi-axis milling operation planning system.

Related projects: IAP P05/06 AMS

Publications and reports: 2003PP001, 2003PP016, 2003PP042, 2003PP068

Scientific staff: B. Lauwers, J.-P. Kruth, G. Kiswanto, P. Lefebvre

2.5.2Tool path planning and generation for rapid protyping of sheet metal forming

Raipd protoyping techniques of sheet metal forming like single point incremantal forming (SPIF) and laser forming are very complex and novel processes. A complex shaped sheet metal part is formed by moving a cylindrical tool (“pen”) or laser beam over the surface. Classical tool path generation strategies for milling, available within a commercial CAD/CAM system, have been studied for use for these forming processes. This CAD/CAM system is also used to generate various tool paths for the experimental investigation of the incremental forming process.

The development of tool path generation strategies dedicated to the incremental forming process is on-going. Specific process characteristics such as thinning of the deformed sheet, spring back behaviour and the part complexity are taken into account.

Related projects: IWT Project 30262 "Semper"

Scientific staff: B. Lauwers, J. Duflou, K.R. Mahajan, H. De Baerdemaeker

2.5.3Collision-free NC programming through the integration of tool path generation, postprocessing and NC-simulation

Collision avoidance is a well-known problem in the NC-programming of multi-axis milling machines. Collisions can occur between part and machine, tool and part or between moving machine components. Generally, they can only be checked by executing some test runs on the machine or by using an NC-simulation program. If collisions are detected, the NC-programmer has to change milling strategy and has to regenerate the tool path.

Advanced NC-programming methods for the generation of collision and error free NC-programs have been developed and during previous years implemented within CAM software for multi-axis milling.

During the last year, the developed concepts have been used to develop dedicated CAM strategies and NC-programming software for the robot based machining of complex shaped products. The research work also included the development of collision avoidance methods. Examples of collision avoidance methods are the selection of an alternative machine axes configuration, change of tool length, change of part position and simple retraction of the tool. Within the same concept the redundancy of complex kinematic structures (e.g. robots) can be used to avoid collisions

Related projects: EU Project no. GRD1-200-25135 Autofett

Publications and reports: 2003PP046

Scientific staff: B. Lauwers, H. De Baerdemaeker, P. Lefebvre

production planning

2.5.4Production planning for sheet metal bending

A topic receiving only limited attention in the sheet metal industry is the production planning problem for the bending process, specifically for air bending. In an attempt to automate the whole process of tool selection and process planning, the usefulness of a well considered production planning algorithm became obvious.

This research focusses on the search for good planning algorithms, applicable for the sheet metal bending process. In a first stage, the objective is to minimise the makespan of all planned orders. Sequence dependent change-over times and very large change-over times in comparison with the actual production times do not allow to use standard planning algorithms available in the literature.

Related projects: Intregrated tool selection and production planning for bent sheet metal parts

Publications and reports: WP2002-47, WP2003-22, WP2003-07, WP2003-21, WP2003-23, WP2003-02, 03EC26

Scientific staff: J. Duflou, D. Cattrysse, J.-P. Kruth, D. Van Oudheusden, Ph. Collin, B. Verlinden

holonic manufacturing systems

2.5.5Multi-agent Coordination and Control Systems

Coordination and control is a research domain whose importance is growing with the increasing penetration of computers, sensors, communication and software in our society and in the artefacts we develop. Just as speech, vision and natural language, coordination and control are intrinsic to human society. Agent technology is considered to be a most promising candidate technology for addressing the challenges in this research domain in a generic manner.

The research team specifically investigates multi-agent manufacturing control, where coordination mechanisms from insect societies are combined with state-of-the-art software engineering methods. The prototype developments use the PROSA (i.e. product-resource-order-staff agents) architecture. Ant colony coordination techniques provide emergent short-term forcasts of the system behavior. Furthermore, the team investigates how coordination and control can be applied across application domains. The team investigates extension of its techniques to multi-site and multi-organisation coordination and control problems - supply chains, virtual enterprises.

The ultimate objective of the research team is to carve out the reusable parts of a multi-agent coordination and control system, and develop what could be called a 'coordination and control operating system' to which the system-specific parts are added as an application.

During 2003, the research team has developed a research platform in Java on which experiments can be performed. This platform emulates the underlying system, which is being controlled, such that speed-up is maximised while accounting for the deliberation time needed by the coordiantion and control system. Furthermore, the team started the development and implementation of its resource coordination technology in a fully distributed version, based on JADE, in the context of the MABE project.

Related projects: GOA/2001/05 - AgCo2, GRD1-2000-25065 - MPA, GRD1-2001-40496 - MABE, IST-2001-65001 - IMS-NoE

Publications and reports: 01PP013, 03PP073

Scientific staff: H. Van Brussel, P. Valckenaers, O. Bochmann, C. Zamfirescu, Hadeli, B. Saint Germain, P. Verstraete

2.5.6Emergent behaviour and self-organisation in complex manufacturing systems

A system exhibits emergence when there are properties at the macro-level that arise from the interactions between the parts at the micro-level. This emergence cannot be understood by taking the system apart and looking at the parts. They can however be studied by looking at each of the parts in the context of the system as a whole. Self-organisation is a process where systems acquire and maintain structure without external interference. This does not exclude data inputs from outside the system as long as these inputs are not control instructions.

The objective of the research is to better understand how to design and/or synthesize adaptive coordination and control systems that use emergence and self-organisation.

These natural multi-agent systems have been taken as an example to develop new distributed manufacturing control systems because they show a lot of potential: the 'simplicity' of the design of the individual components and their interaction mechanism, the automatic 'guidance' to the 'optimal' solution, and the capability to handle dynamic situations. To make more effective decisions, globally relevant information is made locally available via artificial 'pheromones'. This propagated information represents relevant facts. That is to say that this information does not represent the solution but rather the problem world/situation. Current research focuses on emergent forecasting in complex and dynamic systems.

Related projects: GOA/2001/05 - AgCo2

Publications and reports: 2003PP088

Scientific staff: H. Van Brussel, P. Valckenaers, O. Bochmann, Hadeli, B. Saint Germain, P. Verstraete