The sensitive touch
Machines with feelings are the province of science fiction — perhaps fortunately. But advances in sensors mean that machines that have a sense of touch are swiftly becoming a reality.
An EU-funded project aims to develop a 'biomimetic finger', combining robotics, electronics and materials science to make an artificial copy of the human finger that can be connected directly to the central nervous system. This brings closer the prospect of prosthetic hands that can provide the same sort of sensory feedback as the natural senses.
Called NanoBioTact, the project is set to start in the next few months. Project co-ordinator Mike Adams of Birmingham University explained that the project is aiming to exploit recent research on how to integrate signals from synthetic sensors with the nervous system. 'There are quite significant developments in attaching prosthetic limbs to the body and in connecting up this sort of information directly to nerve endings,' he said. 'But current prostheses are equipped with relatively crude force transducers, derived from robotics; they're designed to help you pick up a polystyrene coffee cup without spilling it. What they're not designed for is tactile sensing.'
NanoBioTact aims to correct this by optimising and developing new types of sensors and methods for processing the signals they produce. The project follows another EU study called Cyberhand, which finished recently.
Co-ordinated by the Scuola Superiore Sant'Anna in Pisa, Cyberhand aimed to build a robot hand that could be linked directly to the central nervous system, rather than being controlled by the electrical potentials generated by muscle cells, as with current advanced prostheses.
One of the problems the Cyberhand aimed to solve was that the electrodes through which the prostheses communicates with the nervous system have to be implanted surgically. Because the tissues of the body are constantly growing and being replaced, implanted electrodes can be pushed out of position, or the nerves they need to link to can move away from them.
One of the Cyberhand partners, the Fraunhofer Institute for Biomedical Engineering in St Ingbert, Germany, has developed electrodes made from shape memory alloy, 6cm long and 10µm thick, which can 'home in' on the nerves. Heating these electrodes at various points along their length deforms them subtly, so that they can make tiny movements to push themselves up against the nerves. These developments should allow signals to be sent and received but the Cyberhand is still some way from mimicking the hand itself.
NanoBioTact, which is receiving a €3m (£2m) grant under the EU's Framework 7 R&D programme, brings together a multi-disciplinary group from academia and industry to help close the gap, said Matthew Leach, head of research into touch and appearance at Unilever, one of the project's partners.
'We have neuroscientists who understand the response of the nerves in the finger; sensor technologists and robotics specialists; and materials science and modelling, which is very important because a lot of the subtleties of touch are to do with the way that skin behaves,' he said.
'We also have signal processing specialists, because as we move towards a true biomimetic finger, which requires multiple sensors in the finger just as the eye has multiple sensors in the retina, we need to process all that data into useful and meaningful information.'
The Pisa team is once again involved, looking at the robotics side of the project, while Gothenburg University in Sweden contributes neuroscience expertise. Birmingham University is looking at the materials science aspects and, along with a team from Louvain University in Belgium, at psychophysics, which is concerned with how the brain handles inputs from synthetic sensors. Finite element modelling specialists at Rockfield, a UK-based company, are analysing the mechanics of fingers, while Munich University is heading the signal processing studies.
Signal processing is perhaps the most challenging part of the project, Adams said. 'In humans, you have encapsulated nerve endings which allow you to detect low levels of stresses — they're called mechanoreceptors and they have a certain specific response characteristic. You have many hundreds of these sensors, which send signals through the central and peripheral nervous systems and are interpreted very quickly. If you rub your fingers over a table you are crunching all of that information very fast, in millisecond timescales.'
The NanoBioTact partners are developing a range of nanoscale MEMS (NEMS)-type sensors to detect the same sorts of stresses in the artificial skin of a prosthetic.
'A major aim of the project is to optimise the design of the individual sensors, their spatial arrangement and how they process the data, all based on biomimetic principles,' Adams said. 'We're learning from the way the human body does it — there are strategies for information processing that we are trying to mimic.'
Although prosthetics are an important target for the research, Matthew Leach said some industrial applications could be quite different. For example, although Unilever's involvement springs partly from its expertise in the properties and performance of skin, the company's interest in the project is partly motivated by its product testing programme.
'We do an enormous amount of consumer testing but it's quite slow and you can only test a few products at a time, plus you need to have everything fully safety-cleared before you can test it on consumers,' he said. 'For early stage prototyping, what we're looking at is something to predict the consumer response before the product goes to real people. This technology could allow us to move towards high-throughput screening of new ideas, so only the best test products go out to consumers.'
Using a model of the sense of touch, rather than depending on human response, could be a real advantage, because it allows touch to be isolated from other senses in a way that is not possible for humans. 'It's very easy to distract people,' Leach said.
The look and smell of a product has a major effect on people's perceptions of it and can skew the tactile information — the texture of a product, its viscosity and oiliness and how these combine to form the 'skinfeel' — which is so vital to producers of toiletries and topical treatments.
The project partners aim to deliver the biomimetic finger at the end of the three-year project. 'We're aiming for full delivery of a working prosthesis at the end,' Leach said.