After decades of research, exoskeletons, or wearable robotic limbs, are leaving the lab to be used by medics, the military and industry.
Nick Park’s image of Wallace being catapulted each morning into robotic trousers has brought a smirk to the face of thousands of filmgoers.
But the Wallace and Gromit concept, of wearing a mechanical device to aid mobility, is actually closer to real life than you might imagine. Exoskeletons, which can be used to boost human strength and endurance, are beginning to achieve commercial reality.
While many of the biggest recent advances in robotics have concerned artificial intelligence and enhanced autonomy, hopes are high that this parallel strand of development — a marriage of brain and machine — could soon yield results in a range of applications.
These machines mimic the limbs of the human body to which they are temporarily attached. Instead of muscles, they have powered actuators so they can be stronger and work tirelessly. They are meant to move in step with the desires of the wearer and to be ‘transparent’ in use. And, of course, they must be safe. These are extraordinarily difficult challenges for engineers but, after decades of research and improvements in control and computing power, exoskeletons are now approaching adolescence and are fit to leave the lab.
HAL-5 is likely to be the first to enter the market. It is a battery-powered frame for upper and lower limbs and is to be sold by Cyberdyne of Japan for about £10,000. It has been developed over 10 years by Yoshiyuki Sankai at the University of Tsukuba and would appear to have been styled to appeal to robophiles.
The 24kg full-body version has its own battery pack so that it can work autonomously for up to two hours. The makers claim it can help wearers to stand, sit, walk, climb up and down steps, squat and carry objects up to 40kg heavier than they would be able to unaided. Each HAL (Hybrid Assistive Limb) is custom made to fit the client. Sensors are attached to the wearer’s skin to detect the electromyography (EMG) signals that are sent to the body’s muscles before they move. HAL monitors these signals and its processors decide how the actuators on the artificial limbs should be powered to match the intentions of the wearer.
Although Cyberdyne is already developing a child’s version of HAL, before its adult model has been officially launched, the largest market for the product is likely to be among those at the other end of the age range: the elderly.
The market for assistive exoskeletons for the elderly will inevitably grow as the population of industrialised countries ages, with mechanical aids enhancing lost mobility for the frail. A lower limb exoskeleton may give the wearer enough power to stand up from a sitting position and to walk without fear of stumbling.
Cyberdyne also claimed that a major application for these robot suits will be medical rehabilitation. Research into rehabilitation using exoskeletons is well under way at Salford University, where Prof Darwin Caldwell at the Centre for Robotics and Automation leads a team that has developed several wearable devices, including a lower limb exoskeleton to help people stand and walk. ‘It’s not designed for a walk in the park but it will help prevent secondary medical conditions that can arise through inactivity,’ said Caldwell. ‘Two or three 45-minute sessions a week may be enough to maintain the health of the wearer’s own limbs,’ he added.
The Salford machines have pneumatic actuators, soft cylinders of rubber and nylon that contract like a muscle when air pressure increases. These ‘muscles’ operate at 400Kpa (four bar). ‘Weight is not an issue with them and they are very powerful compared to their weight,’ said Caldwell. ‘One actuator doesn’t weigh much at all but it can lift 300kg. The lower limb exoskeleton weighs less than 10kg.’ A second device simply augments knee strength, locking to enable the wearer to stand.
So far they have been worn only by the Salford researchers because funding has been limited. Making the exoskeletons fit the user has been challenging. ‘The human body doesn’t do exactly what you would expect and it’s difficult getting an exact kinematic match,’ said Caldwell.
For arm rehabilitation Caldwell has built an upper limb device of aluminium and steel weighing just 2kg. It is attached to the user at the elbow by a Velcro strip and has seven degrees of freedom — three at the shoulder, two at the elbow and the rest at the wrist. One idea is that it will help patients to regain neurological and muscular capabilities by helping them exercise.
‘You can train the exoskeleton in a particular set of movements so the patient repeats the exercise exactly,’ said Caldwell. ‘You can know the speeds, forces and velocities exactly. We are working on ways to display the information back to the patient so they can see the progress and make it more entertaining and rewarding for them.’
Another rehabilitation exoskeleton is being developed by Prof Günter Hommel’s team at Berlin University of Technology. ‘One of our major goals is to detect the intention of the user so that the exoskeleton reacts in a helpful way and does not impede freedom of movement,’ said Hommel. As with HAL, sensors on the skin are used to detect the EMG signals that fire human muscles. ‘But these [signals] are very noisy and they change with the user’s temperature and perspiration so we have had to calibrate them continuously, every second,’ said Hommel.
The continuous adjustments appear to have improved the ‘invisibility’ of the powered attachment for the user. ‘In our first trials it was like walking in honey but we have made improvements,’ said Hommel. His project is funded by the university and Germany’s national research board, DRG. He expects to be able to supply an exoskeletal device to a clinic in Ulm for patient use in the next six months and is negotiating with a manufacturer aiming to market the lower limb device.
Just as human bodies have evolved over time, so the systems for automated and robotic therapy change. ‘There is a very wide spectrum of technologies that is being deployed,’ said Prof David Bradley of the University of Abertay, Dundee’s school of computing and creative technologies. ‘Some are very expensive, some low cost, but it’s evolving as physiotherapists and engineers work together.’
His department’s exoskeleton project, started in 1999, has evolved from a system for supplementing walking to a crutch-based design that lifts the foot off the ground, controls the ankles and senses balance. ‘Technically it’s not an exoskeleton any longer,’ said Bradley. ‘Working with the physiotherapist we realised we could achieve the same goal by moving the kinematic structure away from the limb.’ The university is filing patents to protect its innovations.
Divorcing the machine from the human means that Abertay Dundee’s solution is now relatively low cost — and therefore more likely to make the move from the lab to the real world.
An exoskeleton project that didn’t reach the market, however, took place at Newcastle University in the 1990s. A team led by Prof Garth Johnson (whose grandfather, HH Johnson, by coincidence, edited The Engineer in the 1940s) collaborated with an Italian team on an arm exoskeleton for wheelchair users. It was to provide upper limb movement to people with high-level spinal injuries that had rendered them completely paralysed. Two prototypes were built but the concept was not taken up commercially.
The arm was actuated by cables and the Italians, led by Massimo Bergamasco at the Perceptual Robotics Laboratory of Scuola Superiore Sant’Anna in Pisa (SSSP), have continued with this approach to the point where the team is now setting up a commercial company to market its expertise, software and hardware.
Last year Antonio Frisoli, assistant professor of applied mechanics at SSSP, demonstrated its innovative lightweight arm exoskeleton which could play a role in rehabilitation but has already found other applications in virtual reality (VR) work because it has a force feedback capability.
There’s a lot of innovation in it,’ said Frisoli. ‘We have used carbon fibre to make it very light and this helps to give a very high ratio between power and weight. We have designed some special parts to make it almost as moveable as a human arm. It can reach 90 per cent of the arm workspace.’
The portable arm, actuated by cables and powered by electric motors, has been used in two very different VR applications. Fiat has been using it to explore the design of car cockpits. The wearer dons VR goggles to be totally immersed in the car’s interior and then moves his arm, clad in the exoskeleton, to touch the controls. The force feedback completes the illusion and the data acquired from the wearer’s movements shows the designers if their layout is good for human interaction.
The Pisa arm has also gone on an international tour of art galleries, where the public have worn it to ‘touch’ virtual models of classical sculptures, artworks that are too valuable to be exposed to such intimate exploration. In contrast, a remarkably similar upper limb in development at the University of Washington, Seattle, is to be used to measure human arm movement and improve understanding of muscles and their mathematical formulation.
Frisoli’s team in Pisa has spent the last year designing a full-body exoskeleton for the Italian ministry of defence. ‘We have a backpack that will support the spine, two legs and the arms connected. We don’t have a target weight yet. It will lift 100kg so I think we are going to have a weight of 70–80kg,’ he revealed. The limbs will be actuated by cables and the team must perfect a way for the machine to maintain its equilibrium. The first prototype will be ready in a year.
The US military has a long history of interest in exoskeletons and the Defence Advanced Research Project Agency (DARPA) is funding at least two full-body systems. The more public of the two is the Berkeley Lower Extremity Exoskeleton (BLEEX), developed by Homayoon Kazerooni, professor of mechanical engineering and director of University of California Berkeley’s Robotics and Human Engineering Laboratory.
BLEEX 2 is a self-supporting 14kg exoskeleton and backpack that contains the power unit. The machine can reportedly help the wearer, or ‘pilot’, to run faster than 2m/s and carry up to 45kg. More than 40 sensors and hydraulic actuators form a local area network for the exoskeleton and they function much like a human nervous system. The sensors, including some that are embedded within the shoe pads, are constantly providing the central computer with data so that it can adjust the load based upon what the human is doing. The exoskeleton continuously calculates what it needs to do to distribute the weight so that little or no load is imposed on the wearer.
We are taking great pains to make this as practical and robust as possible for the wearer,’ said Kazerooni at an early demonstration. ‘Several engineers around the world are working on motorised exoskeletons that can enhance human strength, but we’ve advanced our design to the point where a “pilot” could strap on the external metal frame and walk in figures of eight around a room. No one else has done that.’
One significant challenge for the researchers was to design a fuel-based power source and actuation system that would provide the energy needed for a long mission. The UC Berkeley researchers initially used an engine that delivers hydraulic power for locomotion and electrical power for the computer. The engine provides the energy needed to power the exoskeleton while being easy to refuel in the field.
The Sarcos Research Corporation of Salt Lake City, Utah, is also developing a full-body exoskeleton for DARPA. It too has hydraulic actuators and an internal combustion engine power pack. The company, run by Prof Stephen Jacobsen, has produced two prototypes which reports suggest can help the wearer to carry 84kg effortlessly.
So it seems that the endearing popular image of Wallace’s walking trousers may indeed be more than a flight of Nick Park’s animation fancy. Exoskeletons could be helping the military, medical and VR communities steal a march on human bodies within the next decade.