Enhancing prosthetic limbs

An artificial muscle being developed using shape memory alloy wires that work like natural muscle fibres promises prosthetic limbs that offer more natural movement for amputees.

Chris Harris, professor of neuroscience at

, is leading cross-disciplinary research to develop the muscle based on a model of how the brain and muscles interact to produce movement.

'We're very stereotyped in the way we move,' said Harris. 'If you reach for an object, you do it on a very smooth trajectory. It's to do with how the neural control is organised to optimise accuracy and speed. The brain sends control signals to a local control circuit in your spinal cord and expects certain things to happen.'

The project's vision is to build prosthetic devices or human-like robotic tools that mimic the stochastic, or noise-generating, properties of real muscles.

'When you generate more force, you also generate more noise [irre- levant signals] proportional to the mean force,' said Harris. 'This is a common biological phenomenon, but is not common in artificial systems. The noise originates in the motor neurons, which innervate (supply nerves to) muscle fibres. As you switch on more of them, you get more noise, which makes actions less accurate. The only way to overcome that is to go slower, like the difference between slowly picking up a pin from a table or just quickly touching the table. Speed and accuracy is a trade-off that is unique to biology, and the brain has evolved to expect that trade-off.

'We'd like to build a neuromorphic muscle, a muscle that behaves like a real muscle with multiple parallel fibres. Technically that's very difficult and it's much easier to use the traditional approach of a motor driven by a control circuit.'

The researchers will explore the properties of nickel-titanium memory alloy wires, which can be extruded until they are very thin and used in parallel. Like muscles, they generate force when a current is applied, but in doing so generate the heat which causes contraction.

'To start with, we'll experiment with quite big, thick wires, around a millimetre in diameter, but I want to go down to much thinner than that,' said Harris. 'I don't know how thin we can go — there are a lot of material questions we need to investigate: strength, conductivity, heat dissipation, how much force you can get out of them — there are a lot of unknowns.'

In real muscle, sarcomeres — the contractile elements of muscle fibres — are in series as well as in parallel. The project will work out whether it is best to have lots of full muscle-length wires in parallel or to have shorter lengths in series in the artificial muscle. Natural muscle fibres also have elastic properties, so an element of elasticity will be incorporated in series with each wire to make it automatically return to its original state.

In nature, only the requisite number of muscle fibres are engaged to produce the required amount of force. 'That's the reason for going to this parallel route,' explained Harris, 'when you generate small forces you'll only switch on a few of these wires.

'I'm very keen that we use this biomimetic approach for a reason rather than copying nature for its own sake. We understand why they're switched on in the order they are — we have a patent pending on the design of a controller that replicates that — but the key is finding the materials to build these parallel contractile elements. We need to ensure the right number of wires contract when a current is applied, relax when you turn off the current and cool quickly enough.'

Muscles only contract; they can pull but never push. In nature, two-way movement is provided by 'antagonist' muscles that pull in the other direction, so this would also be replicated in the artificial muscle.

The three-year, EPSRC-sponsored project starts in October 2009 and consists of three key phases: exploring the use of nickel-titanium wires, developing a controller that drives the wires with optimised signal-to-noise ratio, and detecting signals from the human brain to drive the artificial system. It will bring together three faculties at Plymouth University and the researchers hope

to involve a commercial artificial muscle company.

At the end of the project Harris aims to have a physical proof of concept in the form of a working muscle consisting of several fibres.

'The advantage of our approach is the artificial muscles could be any size you like, just like human muscles,' he said. 'We have very small muscles in the eyes or ears and very big muscles in our legs. It's a trade-off between the amount of force you want to generate and the accuracy you need.'