Robots were once confined to car plants. Now, they are making their way into many diverse areas thanks to improved flexibility, easier programming and reduced costs. David Fowler explains.
To many in UK industry the word robotics still conjurs up visions of lumbering metal giants moving awkwardly around heavy industrial plants.
But robots have been technically sophisticated for quite some time and are being improved constantly. So whether you want a small, inexpensive one for your desktop or workbench, or a big one with more load capacity, there’s plenty of choice.
Ironically, the main limiting factor on the potential of robots is the imagination of humans. Or, as one expert in the field put it: ’We’re aiming to minimise human interaction, and make the machine clever enough to counteract the fact that humans are operating it.’
A number of factors have made robots more accessible. They’re more flexible and easier to program; better integration with vision systems has opened up new applications; and a reduction in costs has brought them within reach of many more users. Whereas once the stereotypical robot application was welding and painting car bodies, now they’re appearing in many diverse areas such as the food industry and in the research laboratory.
Improvements in vision systems also allow them to be used for handling randomly-sized objects, from airport baggage to quarry products. ’A vision system which would have cost £20,000 five years ago is now £2,000. Then it would have needed an expert programmer. Now you just need to be able to use an Excel spreadsheet,’ said Dr Ken Young, chairman of the British Association for Robotics and Automation, and head of the automation research group at Warwick Manufacturing Group. Robot makers and vision system specialists, such as Kuka and DVT or Motoman and Cognex, have been working together to integrate their offerings better.
This wider availability of vision systems is one factor behind a novel application in pharmaceutical giant GlaxoSmithKline’s research and development laboratories.
Alan Stanley, an investigator in the chemistry technology department, works on developing new processes for scientific experiments which when perfected will be used in drug research.
He said: ’My interest is in smaller, six-axis and Scara (selective compliance assembly robot arm) pick-and-place robots to use on the desktop to automate research scientists’ work.’
The robot and machine vision systems had to use off-the-shelf technology, and be modular so that it could be easily plugged in where needed.
In many experiments the researchers need to add a small but precise quantity of a reagent or reagents to a number of test tubes, to investigate the effect of concentration, by adding differing proportions of different chemicals. It’s laborious work and the researchers use arrays of bottles or test tubes in racks, or for smaller amounts, ’microlitre’ plates which contain an array of shallow wells in a 12.5cm x 8.5cm plate.
An emphasis on cost-cutting, and reducing the amount of expensive chemicals used, has led to 96-well arrays giving way to plates containing 384 or even 1,536 wells in the same area. So-called labs on a chip – glass, plastic or polymer etched with microchannels in which chemicals are forced to mix – may have wells in a variety of different positions.
Stanley, working with Warwick, devised an application to automate the dispensing process. ’The vision system checks the plate is the right type, its orientation and finds out where the wells are,’ he said. It then dispenses the correct chemical using a syringe driver. For lab-on-a-chip applications, positional accuracy of up to 200µm could be called for.
Robots used have included Staubli, Epson, Toshiba, Mitsubishi and Kawasaki six-axis and Scara machines. Stanley said: ’We’ve invested a lot of time and effort on the front-end controls. The software was an important aspect.
’In manufacturing, you essentially program the robot once and then use it in that configuration forever. We need repeatability and accuracy, but we also want the software to allow us to vary what we do from experiment to experiment.’ Aitken Scientific collaborated on writing the software to allow the robot to be re-programmed easily from a PC.
Another laborious task commonly facing researchers is the need to weigh each test tube or vial in a rack individually. Stanley is working with Warwick on a more flexible solution to this, and they have developed an adaptable automated weighing station which is in the process of being patented. This will deal with several different types of rack, with and without lids, containing up to 24 vials or 96 tubes. Weighing 24 tubes manually can take a researcher 20 minutes.
’The scientist just has to open the door, put the rack on the table and press go. A single sensor – not machine vision – decides what type of rack it is, in what orientation, and whether it has a lid or not,’ Stanley explained. Having identified how the rack of tubes is arranged it goes on to do the weighing.
Ultimately Stanley wants to address the issue of lab equipment which is used once or twice a week and idle the rest of the time. ’One idea we have, using this vision-and-sensing robot, is to make a robot which can almost configure itself for the next operation. You’d put the module on the bench and it would look around and verify it was in the right place. So it could be a weighing machine today and a dispenser tomorrow.’
He added: ’What we’re trying to do with these systems is remove a lot of user input. If someone puts an object in the wrong place and the robot head crashes into it, you can quickly write off some expensive equipment.’
Flexibility is also the key to a new development, ABB’s Flexpicker, which is finding favour in the food industry. This does not resemble a conventional six-axis robot at all. Instead, it has a circular base about 600mm in diameter mounted overhead, from which descend three arms, meeting at a point like an inverted tripod. The arms are made of carbon fibre.
The machine has a limited operating envelope, but it is extremely effective for picking up objects and moving them from one conveyor to another. The low inertia of the lightweight arms means it can move very quickly – up to 150 cycles a minute with a load of 1kg. A vision system allows it to deal with objects at any orientation.
Typically, it is used to grip an item of food and transfer it, for example, from a conveyor coming out of an oven to a packaging line. ’In one application, eight Flexpickers on a croissant line, with a vision system, can transfer thousands of croissants an hour to a flow wrapping machine,’ said Martin Walder, managing director of ABB Robotics UK. The company has also worked with packaging machine makers to integrate the Flexpicker into their machines.
’The food industry is relatively conservative, and it has taken a while for manufacturers to make big investments in robot automation, but it’s starting to happen,’ said Walder.
Sidebar: Now, robotic teamwork
The motor industry has the longest history of using robots, and development continues even in this relatively mature sector of the market.
Traditionally, while framing vehicles, large jigs have been used to hold the parts in place – for example, pressing the body side while it is being welded to the floor pan and roof panel. Robots now have the capacity to present the simple framing tools themselves – in other words, one robot can hold the components in the correct orientation while another performs the welds, eliminating expensive jigs. This method has been employed on the new BMW 7 series and Renault Megane.
Even greater possibilties have been opened up by linking the robots via a computer network rather than treating them as stand-alone components.
’This means that from a PC anywhere, you can go into a plant and check the health of any machine, download new codes or back up existing programming to a central computer,’ said ABB Robotics UK managing director Martin Walder.
ABB is currently supplying 250 machines for the new Land Rover Discovery, all on an Ethernet link and linked back to ABB’s WebWare software.
’If there’s a revision to the programming code or an upgrade, you can download it to all the machines after the last shift. Previously it would have taken days. And you always have the latest revisions on the central computer, so you can see if there have been any unauthorised changes,’ added Walder.
ABB’s RobotStudio software allows robotic production cells to be modelled in advance. It can download details of the component details and the plant environment from CAD files. It then simulates any of ABB’s robots, modelling reach, and predicting loadings and cycle time.
’It also cuts down the critical path in setting up production,’ said Walder. ’You can program a whole project offline, before the robot or tooling have been installed or even built, and download the code to the robot controller when it’s installed for fine tuning.’