RPD makes itself heard

The concept of the home office is now commonplace, with broadband and the internet making it a perfectly ordinary lifestyle for millions. Now the home factory – a production line in your house that could manufacture lots of different products, virtually for free – could be just around the corner. Adrian Bowyer, a senior lecturer in mechanical engineering at Bath University’s Centre for Biomimetics – is leader of a project called Replicating Rapid Prototyper, or RepRap. One element of the project is rapid prototyping – the automatic construction of physical objects from 3D software models. This is a usually beyond the resources of most individuals. But underneath this is another extraordinary idea: self-replication. A rapid prototyper can build many different objects. So why not make one that can build a copy of itself? If you can do that once, you can do it again and again. Result: the cost falls to virtually nothing – just the raw materials – and RepRap machines become available to millions. If it works, it will be the best application yet of one of the most beguiling ideas in the whole of technology: self-replicating machines. Bowyer says the ultimate aim of ‘pure self reproducing machines’ is to create systems that can make exact copies of themselves with no human intervention – apart from initial provision of raw materials. However, he says that approach fails to distinguish between the inherent abilities and drawbacks of human beings and machines. People, he says, find it difficult to carve a block of plastic to an accuracy of 0.1mm – which is easy for a computer. Putting carved pieces together – particularly if they can only fit one way – is easy for us but harder for a machine. “I want to produce machines that do what they're good at, while we help by doing what we're good at,” he says. “What that means is making a machine that can make all its parts. Then a person can put them together.” Clearly, making machines that do nothing but copy themselves would be somewhat pointless. So the aim is to create machines with a self-reproducing capacity, but which can also make all sorts of other devices, from coat hooks to cat flaps. Almost anything is possible, within the size limitations of the systems currently being developed – currently a 300mm cube. The other key element of his strategy is to give away the RepRap for free – the whole system’s electronic design, software – everything needed to build it – will be available from the Web. The logic is that if you create a system that can copy itself, and anyone can have one, you have the potential for an exponential increase in the numbers of such systems being produced. This leads to a similar fall in cost – of both the self-reproducing machines and the objects they can make. It may sound absurdly generous but the economics are driven by logic: once a system can copy itself (with a bit of help) it costs very little other than the raw materials to produce unlimited numbers of it, which makes its monetary value approach zero. “It is potentially extremely wealth creating but itself is worth nothing,” Bowyer says. There is also a moral dimension, in that he did not want such a system to be under the control of any individual, company, or government. “If you have a powerful technology, a good way to make bad things happen is for only some people to have access to it.” One requirement being asked of anyone building a RepRap is that they make two other machines for other people. The team have set themselves a deadline for distributing the first machines by 2008 but Bowyer is quietly confident they can beat this. He is not in complete control of the project – different teams of people are working on prototypes worldwide, some of whom he has never met. Most recently, in September, a RepRap machine developed in Vienna by Vik Olliver, succeeded in producing the first part for itself (see http://staff.bath.ac.uk/ensab/replicator/). First generation machines will produce only plastic products, but the team is already planning a second generation that will handle low melting point alloys, enabling RepRap systems to deal with electrical conductors and hence produce a combined electrical and mechanical object. “These second stage machines will have a deposition head in them that works directly on the alloy, so you will be able to produce electrical circuitry. And people are already working to develop inkjet printers that can print semiconductors on plastic sheets. Once RepRap machines are established, there is nothing to stop them creating a semiconductor print head.” But even with plastic-only output, Bowyer believes the potential is huge. “I cite plastic coat hooks as an example of what could be produced, which sounds completely trivial, but an economist has told me the worldwide market for them is much larger than for massive objects like gas turbines.” Clearly, it will require a change in mind set for all of us to think about making our own small plastic objects. But if it takes off and RepRap machines cost almost nothing – and the crucial requirement for that is their capacity for self-reproducing – it could happen. People will get a RepRap machine in the same way as they buy a conventional printer today, for the same price, say £100. Then they simply download designs for objects from the Web, or create new ones themselves using free 3D modelling software, ‘print’ them out, and the home factory is born. The self-replication concept can apply not only to the RepRap machines but to the raw materials. In future, Bowyer is hoping to be able to use a polymer called polylactic acid, which can be made from fermentation from starch using potatoes, or maize. “That means if you've got a few tens of square metres of ground you can have a supply of raw material that copies itself. And of course, the RepRap can also make a fermenter. Also the plastic is fully biodegradable so it can go on a compost heap, and the result is you have immediately a local recycling route. That makes it extremely benign ecologically.” Further developments could include using the RepRap to make moulds, and from them produce anything that is mouldable, with materials like epoxy, concrete, plaster of Paris or ceramics. For some observers, there are two Holy Grails of future technology and they both involve self replication: physical machines of some kind that can copy themselves, and software programs that can learn and create better versions of themselves. Why are these so significant? Because once you reach these points, there will be an exponential increase in what is achieved – and we can sit back and watch. Adrian Bowyer Email: A.Bowyer@bath.ac.uk Tel: 01225-385453 Hod Lipson Email: hl274@cornell.edu The idea of a machine that can copy itself goes back way before the days of electronic computing, at least to the 1870s, when it featured in Samuel Butler’s novel Erewhon. Computing pioneer John von Neumann described a Universal Constructor, which is a self-replicating machine in a cellular automaton environment. Essentially, he demonstrated the fundamental logical underpinning required for self reproduction. His constructor has 29 possible states, allowing signals to be sent and logical operations to be carried out, and a 'tape' of cells encodes the sequence of actions to be performed by the machine. Using a writing head, the machine can generate (by printing out ) a new pattern of cells, allowing it to make a complete copy of itself, and the tape. In the 1950s, scientist Lionel Penrose and his son Roger – later to become one of the UK’s leading scientists – built a system of wooden cutouts, in several shapes, that could fit together into compound parts. Placed in a tray and shaken, they assembled into specific patterns that would also repeat over ‘generations’. It was effectively a mechanical self-reproducing system. More recently, Matt Moses at the University of New Mexico has devised a system comprised of Lego-like bricks to make a three-axis manipulator that can assemble a duplicate of itself. However, the machine cannot fabricate its own plastic components and must be controlled by an external entity. To be fully self-replicating, admits Moses, the device would have to control its own actions autonomously and possess the instructions necessary for carrying out its duplication. Hod Lipson of Cornell University has made a tower of cubes attached to each other by magnets, which can be energised or not. Called ‘molecubes, they can rotate and manipulate other cubes in their surroundings, and by doing that produce a copy of the original tower. The results have been intriguing.