Divide and conquer

Citation

Loughlin, C. (2007), "Divide and conquer", Industrial Robot, Vol. 34 No. 6. https://doi.org/10.1108/ir.2007.04934faa.001

Divide and conquer

As I write this, I have just returned from attending the CLAWAR 2007 Conference (Climbing and Walking Robots, Singapore, 16-18 July 2007) and one of the highlights of the conference for me personally was being able to present the “Industrial Robot Award: for practical innovation in the field of robotics”.

This year the award went to a group Centre for Automated and Robotic NDT, London South Bank University, UK for their paper “Development of a climbing robot for inspection of long weld lines”.

Welding is one of the earliest applications for robots and is still one of the most common, and yet it is also one that continues to generate a significant number of new technical developments and innovations.

Welding together of cars and other fabrications as they roll down, a production line is now somewhat routine. The robots are firmly bolted to the shop floor and they perform the same welding operations time after time. When you are making cars then this is exactly what you want, each car should be the same as the one before it and the last thing you want is variability.

The real challenge for robot welding these days is not the automotive industry but instead lies in the welding together of large structures such as buildings, bridges and ships where virtually every weld is unique and where the sheer size of the operation implies unavoidable variation in part location. You can forget about sub-millimeter accuracies and absolute positioning to a few thou. As soon as you start talking big, everything moves around as thermal and gravitational bending forces come into the equation.

The challenge today is to take the robot welder to the workpiece and not the other way round. The robot must be able to position itself at the correct location and be able to weld horizontally, vertical, overhead and all angles in between. Another added difficulty is that because each weld is effectively unique you cannot apply normal statistical sampling techniques to monitor weld quality. Each section of weld needs to be inspected to verify its quality, and usually this inspection needs to be done as the welding continues so that the operation can be halted if problems occur before too much damage occurs and too much rework is created.

What appealed to me and the other judges about the above paper was that the various tasks of welding and inspection had been broken down and shared between multiple robots. In this case, three robots are used. A utility robot that basically carries the reel of welding wire, the welding robot itself and finally the NDT robot that checks the quality of the welds.

By splitting the tasks in this way, the size and weight of each unit is reduced and because each unit is free to move relative to the other units the overall flexibility of the system is greatly increased. This all means that it will be able to perform a much higher percentage of welding tasks as compared to a larger, heavier robot that attempted to combine all operations in a single unit.

This approach is nothing new because it mimics the way that manual welding has been done for ages. The welding power supply and wire feed is in one box (or sometimes two), the welder holds the torch and the weld inspector follows behind sometime after. What is new however is that this tried and tested technique is now being automated.

The London South Bank University project is at an early stage of development, and no doubt numerous problems remain to be discovered and overcome, but it does demonstrate the sort of lateral thinking that we need to embrace if welding robots are to detach themselves from the factory floor and tackle the great many opportunities that await.

A revised and updated version of the paper will be published in a future issue of our journal and I look forward to seeing how the project progresses.

Clive Loughlin