University of BristolTouch to transform with tactile robotics

Touch to transform with tactile robotics

University of Bristol

3D-printable fingertips could replace manual labour and assist with caring

What separates robots from being “human” is a frequently asked question in the age of artificial intelligence. At present, the answers are numerous. One human ability that researchers believe could benefit society, if successfully emulated in machines, is the use of our hands.

With this goal in mind, researchers at the University of Bristol are developing groundbreaking technologies to improve robot dexterity – specifically, by giving them “fingertips” inspired by the human hand.

The project, backed by a £1 million Leverhulme Leadership Award grant over five years, aims to build “tactile robots” that are able to judge and interact with their surroundings to the same level of dexterity and robustness as human touch.

So how important is our sense of touch? According to Nathan Lepora, professor of robotics and artificial intelligence at the University of Bristol, our hands are “almost useless” without it. “If you anaesthetise someone’s hand so they just use their eyes, it’s very difficult to pick things up,” he explains. “As for more complex tasks such as using a tool or writing, we’re completely dependent on that feedback between hand and brain of how the items feel.”

As the principal investigator for the tactile robotics group, Professor Lepora’s work combines robotics and neuroscience, as well as psychology, to reproduce the human sense of touch in robots, which he believes could “transform the economy and our society”.

The project has already delivered results, including a working biomimetic fingertip that he describes as a “printed skin” for robot hands. It is the group’s hope that TacTip, as it is called, can be downloaded and built by anyone with a 3D printer.

The idea has drawn interest from industry, and Professor Lepora’s group is partnering with a number of commercial companies, from Ultraleap, a Bristol-based start-up, to Google’s DeepMind.

The benefit of working with these organisations, says Professor Lepora, is that they have “core insight” into how to use the technology. For example, DeepMind is a world leader in algorithms that learn from reward, which is how humans learn to use their hands.

Robots are already having an impact on manual labour around the world, with factories and delivery companies such as Amazon utilising machine intelligence. But there are still opportunities for automation, Professor Lepora believes. Packaging in delivery warehouses “remains a challenge for machines”, for example, on account of the delicate handling and complex dexterity required. “Putting stuff in boxes is the hardest part for warehouses using robots, because it requires an understanding of the weight and feel of the item,” he explains. Recycling plants could also make use of more intelligence and sensory machine hands because “a lot of recycling is currently wasted, as it’s so laborious for humans to sort out their rubbish”.

From an economic point of view, it is these menial tasks that will most likely be impacted by the development of robot fingertips, and this does “present legitimate concerns about job losses”, Professor Lepora acknowledges. He believes that society will “adapt with the changes”, however, and that “AI will bring more positive impacts than bad”.

A conscious driving force for researchers at Bristol and its collaborating universities is to cultivate uses of the technology that will benefit society on a local level – and a partnership between Bristol researchers and the University of Pisa aims to address just that.

Antonio Bicchi is chair of robotics at the University of Pisa and senior researcher at the Italian Institute of Technology. He is widely regarded as a global leading expert in the field. His research group has already developed an anthropomorphic “soft” robot hand, SoftHand, and Professor Lepora describes Professor Bicchi as “an inspiration”.

What makes SoftHand unique is that it is the first robot hand designed to replicate how the bones and tendons work together in human hands, which enables the hand to open and close in a more “natural”, human-like way. According to Professor Bicchi, the inspiration for the design came out of a “frustration that none of the designs that we had made in the robotics community was having an impact on real life”.

“We are working with Bristol because tactile sensing has for a long time been lagging behind the other senses,” he explains. “Robotic vision, for example, is much more advanced, but the sense of touch is much more complex. Our skin has tens of receptors in every square centimetre, and replicating the same structure is pretty much impossible. So you need an idea of how to extract the complexity into something simpler but tangible.

“Now we are building hands that are simple to use, so that an amputee can move more naturally, or a robot that uses this hand can be programmed easily.”

A great deal of Professor Bicchi’s time is currently spent working with patients at a hospital in Vienna, where his robotic hand has already been trialled for use by stroke patients without full working use of their limbs.

Bristol’s TacTip and Pisa’s SoftHand have been proven to work together seamlessly, and the product has been demonstrated at conferences to robotics stakeholders around the world. Now, the researchers intend to continue working together to create a fully sensorised hand across all fingers.

Professor Bicchi says there is still “a lot of work left just to understand the basics of intelligence involved in object manipulation, but the technology that Bristol has holds promise, and that is very positive for us”.

His group aims to start teaching the tactile hand to move objects around in its grasp, for example – a skill that could bring experts a step closer towards developing robots to assist with caring jobs. The fingertip sensor can already learn to type using a Braille keyboard, according to Professor Lepora, “which gives a demonstration like the kinds of things humans do with touch”.

A proud moment for Professor Bicchi came when a friend and medical patient sent him a video of her “stroking her baby’s head with her new hand”.

“These projects are very ambitious – to recreate the human hand is a moonshot,” he says. “But you can imagine, we have worked on robotics for so long – seeing it in real life now with patients who use the technology every day, it’s the reason why we persist.”

Professor Lepora agrees. “Robot hands with human-like dexterity could improve our lives in so many ways, from doing menial work in factories to assisting carers with handling patients. It’s really important that society make the most of this opportunity.”

Find out more about the University of Bristol.

Brought to you by