The robot, once the preserve of large industry, could be coming to the shopfloors of SMEs to perform light tasks alongside staff via 'intuitive' programming. Samantha Pears reports.

Some of


's leading research institutes, universities and automation technology companies are planning to bring industrial robotics to the masses. The masses in this case are


's 200,000-plus smaller manufacturing firms, who will have access to intuitive, affordable self-assembly 'light' robots if the EU's

SMErobot project

within the 6th Framework Programme project is successful.

Industrial robots are commonplace in high-volume manufacturers. Almost a million are in service worldwide, and their numbers are on the increase. However, robots are scarce in SMEs due to their cost and complexity. The SMErobot project aims to rectify this.

The term light relates to the construction of the robot. It will weigh around 20kg as opposed to an industrial robot that weighs nearer two tonnes. Light robots are aimed at lightweight jobs, such as arc welding and machining, where an employee now uses a hand tool. However, the robot will not be limited to one or two uses within an application. When woodworking, for example, it might be employed to trim, dress, mill, glue or shape.

The robot will be used in situations where it is not so much the work but the environment that is the problem. The solution? 'Robotising the operation. Removing the employee from the arduous environment, but also removing the health hazard,' said Dr Peter Haigh from UK-based project partners Casting Technology International (CTI).

The aim of the EU-appointed team is to create a light robot that not only performs tasks but takes over a role, such as a repetitive job, that could lead to health problems like vibration white finger (VBW).

Light robots will be modular in design, which will ensure a wide scope of applications, be it for processing wood, metal or ceramics or lifting. And it is hoped that the final result will also be light on the SME's purse strings — the new robot will cost a third of the amount of a conventional system. A robot working in a typical foundry might cost upwards of €50,000 (£34,000), whereas the light robot should be less than €20,000 (£14,000).

A team of leading research institutes, universities and robot manufacturers, including ABB and Kuka, have three main goals. The first of these is to design a robot capable of understanding human instructions. They hope to achieve this by developing a combination of existing devices and methods to create intuitive instruction paradigms. The team wants to produce robots that understand speech and human gestures as well as other automatically generated instructions that will ultimately limit the programming effort.

The final objective in intuitive programming is to incorporate mass-market handheld input devices that require simple human instructions rather than complicated CAD-designed, job-specific programming. Such paradigms exist — programming by demonstration (PbD) is one example — but are not integrated into shop-floor robots.

Programming usually requires much effort and increases the life-costs of the unit because the robot is programmed to carry out a specific task. Altering the robot's work involves hiring the services of an experienced and expensive programmer.

Using established programming methods, what is normally a routine task for a robotic programmer might take the average SME employee 40 hours or more to achieve. The researchers will develop a range of programming methods that are as simple as asking a colleague to perform a task. Haigh explained: 'The success of the project hinges on discovering a new way of programming. Robots cost a lot of money but then SMEs need a CAD person, who is both expensive and not easily accessible. Costs can become mind-blowing.'

The consortium has novel ideas of how to make fast, easy and intuitive calibration of CAD data to cell equipment, tools and work objects possible without having to incorporate complicated measuring data. These might include the measurement of the rotation of a robotic arm at any given time for any given task.

With the new system the robot operator doesn't need to have any knowledge of CAD systems or process model representations. In simple terms a robot hand may hold a plotting device and the user will be able to move the robot hand freely to plot a cutting or shaping path, for example. The robot would then remember the co-ordinates of this movement and repeat its task.

Once intuitive programming is fully realised, the project leaders will create a style guide for anyone developing robot interfaces. This will illustrate how to design multi-modal interfaces based on voice, gesture or manual guidance, so that the robot can interact 'naturally' with humans.

The robot's human interaction will extend beyond intuitive programming, though. Rather than being secluded behind safety fences, most of the new robots will share their space with employees. Currently, robots operate behind fences to ensure operator safety, because of their high-energy motion and the risk of software faults. The challenge for the team is to create systems that are 'intrinsically safe by the laws of physics'.

It hopes to achieve this by radically decreasing the moving mass of today's robots while maintaining performance. The research activities will focus on fulfilling low-inertia and high-performance criteria. So far there has been limited practical success. However, the team proposes to explore new materials, manufacturing processes, structures and kinematical ideas to produce space-sharing robots.

The team will also develop a 'virtual curtain' where the robot's workspace is protected by sensors. Usually, when an object passes the virtual curtain a binary signal brings the robot to a complete halt. However, the new system will be able to monitor when a human enters the robot's workspace and instead of stopping altogether the robot's movements will be adapted from productive, fast speeds to safe, slow and compliant motions. This will call for not only advanced, sophisticated sensing for tracking body motion but also off-the-shelf and low-tech sensors to keep costs low.

The team's third ambitious challenge is to create a 'plug-and-produce' robot. The idea is that the SME employee can assemble and configure the robot and have it working within three days of taking delivery.

To achieve this, protocols and software interfaces will have to be standardised to allow automatic configuration of all the robot's components. Grippers, tools, sensors and feeding devices will have to be automatically interfaced to the cell. Once these protocols are developed, the consortium wants to set European and worldwide standards for them and all plug-and-produce mechanisms.

Although the light robot project is four years long, CTI aims to demonstrate a life-sized working robot within its foundry within the next 12 months.