Point of Contact: Providing Robotics with a Human Touch
08 September 2023
Caption 1: A robot hand using synaptic transistors (courtesy of the University of Glasgow)
In the past robots were bulky, rapidly-moving pieces of machinery that had very little subtlety too them. Isolated from people (for safety reasons) they were just programmed to carry out the same automated assembly tasks day-in/day-out. Our perception of robots is changing though, as are the ways they are being used and the physical forms that they take.
This is calling for cutting-edge advancements in sensor technology, nanomaterials and artificial intelligence (AI).With robots no longer being restricted solely to heavy duty activities, but actually starting to interact more closely with us and their surroundings, their manual dexterity is coming into question. Many of the services that we now want them to provide will depend on them dealing with fragile objects. More effective handling mechanisms are therefore needed. One on side, grips must be secure enough to stop items from slipping out of robots’ grasp, but on the other side they must have the delicacy to prevent breakages occurring or the risk of possible injury.
Basic principles
Finding ways to enhance robots’ manipulatory skills has been something that’s preoccupied engineers for quite a while, but their increasing shift into our everyday environments is adding ever greater impetus for finding applicable solutions. Through the efforts currently being made at various locations, research staff (both in academia and the private sector) have been making huge steps forward in that respect.
Until this stage, equipment has had to rely on haptic feedback (from force, torque or pressure sensors) to determine contact with an object - but this has proved inadequate when a gentle approach is required. Robot gripping claws don’t have sufficient deftness in their tactile movements. That’s not only due to their hard external surfaces, as the resolution of the sensors utilised presents a challenge too. On top of this there’s also the latency aspect - with data from sensing elements needing to be sent back from processing, then the appropriate response being decided upon and the relevant motors/actuators triggered. In the time taken for this to happen, the item may have been dropped or crushed completely.
Possible scope
A plethora of different sectors could be transformed through access to nimbler technology. By increasing their dexterity and making them less clumsy, the breadth of tasks that could be executed would be far greater - thus alleviating acute staffing shortages in certain sectors. Much more intricate assembly on factory production lines could be carried out. It could similarly be invaluable in relation to the cobots already starting to work alongside humans. In a healthcare context, it would be pivotal in automating a large proportion of clinical procedures. There should certainly be benefits in terms of assisted living for the elderly and infirmed too. Smart agriculture would likewise gain for robotic deployments of this kind (for fruit picking, crop tending and suchlike).
Progress made
Right now, a diverse array of robotic touch projects are being conducted around the globe. Here are a few of the higher profile ones.
Pioneering work being done at the University of Glasgow is focused on creation of an experimental skin-like structure made from a flexible plastic composite into which ‘synaptic transistors’ have been embedded. By emulating the neural pathways found in human brains, robotic equipment will be able to deal with external stimuli at source. A more sophisticated, nuanced arrangement will result - calling for less data communication bandwidth and central processing resource, with faster turnaround times and better responsiveness being derived. The current prototype comprises 168 of these synaptic transistors, fabricated from zinc-oxide nanowires.
Caption 2: Use of Tactaxis magnetic soft sensing from Melexis
Across the Atlantic, Caltech boffins are also developing an artificial skin technology for robotic implementation. This features a gelatinous hydrogel into which micron-scale sensors printed on silver nanoparticle wires are placed. Here the concept being explored is of a slightly different nature however - with a human operative receiving feedback from the robot arm’s sensors applied directly to their own skin (through a set of forearm-attached electrodes). Among the potential uses for this will be directing remote surgical procedures, as well as the undertaking of work in hazardous environments (such as in nuclear reactors, or where dangerous contaminants are present).
When it comes to commercialising robotic touch, Melexis seems to be the company nearest to doing so. Staff at its R&D centre in Bevaix, Switzerland, have been investigating this area for several years. Prototypes that the company has demonstrated using its proprietary Tactaxis magnetic sensor technology allow high-resolution 3D force vector soft sensing. Accompanying the sensor is a magnet which has been embedded into an elastomer material (thus providing a malleable contact interface). The deformation caused by an applied force may be precisely determined via the changes to magnetic field witnessed. Thanks to the gradiometric approach employed, the multi-axis sensing device is immune to the presence of stray magnetic fields, thereby safeguarding against the prospect of measurement errors occurring. The team’s objective now is to get the solution’s footprint dimensions down to 10mm × 10mm - so that it is better aligned with the size of a human fingertip.
There are also research institutes looking at the practicalities of mass production. Earlier this summer a team from the Technical University of Munich (TUM) announced the development of a process that in the not to distant future could enable the volume manufacturing of soft sensors via conventional 3D printing equipment. It would then be possible to wrap the stretchable, skin-like material produced around robot parts involved in fragile object manipulation - effectively making such functionality simple to implement, regardless of the shape or size of the parts involved. The upshot of this would be that much greater economies of scale could be benefited from, avoiding the need for making different bespoke shapes for each robot hardware configuration.
Among the other places where work is underway is the Bristol Robotics Laboratory. The angle its researchers are taking is how a sensory artificial skin could be advantageous to those with prostheses. By adding tiny pin-sized sensors (almost analogous to the papillae nerve endings found on human tongues) to the surface of a prosthetic hands, there is the prospect of touch-like capabilities being enabled.
At the University of Tokyo, things are being taken even further. The ground-breaking biohybrid robotics studies that have been embarked upon there involve the digits of a robotic hand having organic cell cultures (a couple of mms thick) grown on them. This is being achieved through the application of a mixture of collagen, dermal fibroblasts and keratinocyte cells. The argument for growing a skin onto mechanical apparatus is that it will then provide an optimal fit, better aligned with any bending movements being actioned - whereas for a sheet cut-to-size there will probably be some unavoidable tolerance compromises. In addition, there may even be the possibility of self-healing properties to factor in (thus extending longevity). At this point it really starts to sound a lot like something out of a James Cameron movie though - don’t you think?
