EC: Project G1RD - 2000 - 00461
Chris Peters*, Keith Herman** Mario Sack*** and Stephen Mulligan*
* TWI Ltd, Granta Park, Gt Abington, Cambridge CB1 6AL, England
** Caterpillar Belgium S.A. 1 Avenue des Etat-Unis, B-6041 Gosselies. Belgium
*** Institut Fabrikbetrieb und -automatisierung, Division of automation, Sandtorstrasse 22, 39106 Magdeburg, Germany
Paper presented at International Colloquium on Autonomous and Mobile Systems, 24-26 June 2002, Magdeburg, Germany
The ability to automate welding activity in an environment producing relatively small numbers of customised products has potentially far-reaching and wide-ranging benefits for the fabrication industry. These benefits will arise through being able to meet a customer's exact requirements, at an acceptable cost and time scale and in improving working conditions for employees.
The NOMAD project aims to develop an autonomous, flexible, robotic welding system capable of fabricating small batches of customised products as easily and quickly as large multiples.
The system will embody the following innovative aspects:
- Use of manufacturing simulation for automated process planning and real-time system monitoring.
- Autonomous robot navigation for high accuracy positioning of a 6-axis robot arm.
- Design and build of an industrially rugged Robot Transport Vehicle with the attributes required for the welding tasks.
- Specially developed welding consumables, welding procedures and sensor systems designed to allow 'all positional' robotic welding with a degree of control and dexterity unmatched by current systems.
This paper describes the route being taken in the development and integration of technologies allowing replacement of the assembly line based approach to welded fabrication with automated, constraints free manufacturing. The system will utilise autonomous, mobile robots to enable the production of high quality, customised products as easily and quickly as large multiples.
There are many influences affecting the ability of large-scale European fabrication industries from remaining competitive. The needs and expectations of customers have shifted from mass-produced 'standard' products to customised products that meet their exact requirements and specification. There is also a drive for shorter lead-times from initial concept to the final fabrication of a product.
Increased price competition from fabrication companies in low cost-base geographical areas around the world as well as a reduction in the number of skilled and trained personnel, particularly welders, are also significant factors.
Traditional European manufacturing companies struggle to respond to the demands imposed by these trends for several reasons. Typically, large-scale fabrications, such as earthmoving equipment and pedestrian bridges, are produced using either manual welding or dedicated automatic (Robotic) welding processes. Manual welding is highly flexible in terms of adaptation to size and shape but is characterised by high cost (labour), low production rate and variable quality. Dedicated automated processes are capable of producing high and consistent quality with high production rates but lack the flexibility required for multiple product types.
This paper describes a route being taken by the NOMAD project to develop manufacturing systems utilising autonomous, mobile robots to enable high quality, customised large fabrications to be produced as easily and quickly as large multiples.
NOMAD - Autonomous Manufacture of Large Steel Fabrications is a European Community Framework V project (EC Contract G1RD-CT-2000-00461) under the Competitive and Sustainable Growth program.
Automated welding using robots has brought an increase in welding productivity and quality to the fabrication industry. The problem has been that current technology has not allowed robotic welding to be applied efficiently to large-scale, low-volume fabrications. Expensive and complicated fixturing, expensive work-piece positioners, and long programming times are among the constraints that have prevented it from replacing manual welding.
In this project the approach is to remove constraints from the process and minimise the time from initial design to final product. To accomplish this goal, a new approach is being taken to robot welding. See the pictogram showing the concept of the NOMAD project.
NOMAD will use a 6-axis robot arm mounted on an autonomous Robot Transport Vehicle (RTV) that will allow the robot to be brought to the part. The structure to be welded will be placed above floor level on a platform so that as the RTV moves around the structure the robot arm can access all the joints to be welded. This approach also will allow for parts to be tack welded and robot welded in the same location which, will reduce the need for transport of these large structures between workstations. For maximum flexibility it is intended that all essential equipment for welding will be carried by the RTV. This includes welding power source, shielding gas, welding electrode, and ancillary equipment.
Successful guidance of the autonomous vehicle will require accurate identification of the position of the robot arm base relative to the welding positions on the structure. This will be achieved with the use of an accurate vision based sensor system that will detect the product type, location and orientation of both the structure and the robot. The figure on the right shows pictures taken during development work on the vision system's part locating algorithms. The top picture is a raw image, the middle picture is an image generated by simulation, and the bottom picture is a superposition of the two images. Information on the location of the robot and the structure will be passed to specially adapted robot simulation software, which will calibrate the simulation model. After calibration is complete, the simulation system will calculate the motion path for the RTV and robot arm based on the location and orientation of the parts in the cell.
Although it is believed that the RTV will reach its destination with unprecedented accuracy, it is not foreseen that the accuracy of the welding torch position will be within that required for welding. Therefore local sensing in the form of touch sensing, laser-based vision and through-the-arc seam tracking will be used to guide the robot arm during welding. It is believed that the combination of global and local sensing and guidance will provide sufficient precision for welding.
Long programming times are a major consideration affecting the use of robots for welding large sized, low-volume fabrications. Even with the use of current off-line programming software packages, the time required can still be too long. The NOMAD project intends to automate the task of robot programming. Welding joint information on weld joint location and weld size will be imported into the robot simulation software along with the geometric information. Not only will the simulation software be used for programming, it will also be used to monitor and control events during the operation.
Automated programming of robot arc welding is not possible if the simulation system does not have intelligence on the welding process itself. In this project, the robot simulation software will retrieve welding intelligence from a searchable database of welding procedures, which will contain information on the relevant welding parameters for joints in the various welding positions.
Development of welding consumables capable of producing high quality welds in all positions will also be undertaken within the NOMAD project. Consumable development will concentrate on producing flexible and tolerant consumables capable of accommodating a variety of joint types, in all welding positions. The consumable development is closely linked with welding process development so that the welding consumable and the power supply compliment each other to achieve optimum welding quality. The results of welding procedure and consumable development will be used to populate the welding database.
The ultimate goal of the project is to build a demonstration cell into which unique pre-assembled structures can be loaded and subsequently welded in a manner which is as efficient as a large volume production cell of today.
One of the major challenges within the project is expected to be the development of a vision system that is capable of determining the location and orientation of a structure. The vision system will have to be capable of sufficient accuracy to allow for automated RTV path planning, robot weld programming, and finally, welding of the structure. The system will also need to be sufficiently robust to accommodate emissions from the arc welding process and the possibility of reduced visibility created by fume.
The RTV design represented will also be critical to the functionality of the overall system. The successful design of RTV will be the one that meets both the technical needs of navigation within the cell whilst coping with the size, weight, and logistics of all the components required for welding.
The greatest overall challenge for the project is expected to be in the effective integration of the many subsystems into a single functioning system.
The project partners
The partnership consists of a balanced mix of suppliers, technology enablers and end users. The partners involved are: Caterpillar Belgium S.A., a manufacturer of earth moving equipment; Delfoi OY (Finland), an applications integrator of simulation software; Robosoft SA (France), a manufacturer of robotic platforms and autonomous vehicles; TWI LTD (UK), a research organisation specialising in joining technologies and automation; Fraunhofer IFF (Germany), a research organisation for automation and robotics; Reis Robotics (Germany), a manufacturer of industrial robots and automated production systems; ESAB AB (Sweden), a manufacturer and supplier of welding equipment and consumables; and Nusteel Structures LTD (UK),a manufacturer of steel bridges and structures.