Image courtesy of Charlotte Leakey.
Our world is being increasingly defined by a series of ones and zeroes. From the smallest phone gyroscope capable of detecting body movement to metal detectors at the airport to automatic PCR machines that test for the SARS-COV-2 virus in a matter of hours, the technology that acquires and transmits this data has made human lives much easier. Over the last two decades, practical artificial intelligence (AI) usage has led to expanding roles for computers in detecting danger, predicting scores, and advising outcomes in fields ranging from security to medicine—often beyond the scope of its original creation and with limited human intervention. Of course, popular debate and science fiction warns about how AI may eventually replace humans across many fields and make human effort obsolete. However, Wei Gao, a professor of Medical Engineering at the California Institute of Technology, sees the advancement of AI as an opportunity rather than a threat.
Gao received his bachelor’s degree in mechanical engineering and his master’s degree in chemical engineering at the University of California, San Diego. He completed a postdoctoral fellowship in electrical engineering at the University of California, Berkeley. Because of his diverse educational background, he pursued the creation of robots imbued with functionalities beyond traditional ones. “In our minds, robots are industrial-level, capable of doing dangerous tasks in agriculture, defense, and space exploration because they can move and perform repetitive tasks, but we are thinking about the future. How can we build a better robot [by giving it] better functionalities and making it smarter? How can we give it more powerful sensing capabilities?” Gao pondered.
Inspired by human skin’s ability to detect temperature, touch, texture, and even certain chemicals, for example, an allergic skin response after rolling around in the grass, Gao set out to develop ‘electronic skin.’ In the past, researchers have developed robots to detect and respond to physical parameters such as temperature and pressure. However, they were unwieldy and impractical—not much more than a thermometer on a remote-controllable stick. Moreover, Gao wanted to diversify the sensors in his robot so that the electronic skin could have an even greater range of sensory capabilities than human skin. In particular, he hoped to detect infectious pathogens for medical applications, neurotoxins or bomb debris for security purposes, and chemical pesticides for agricultural uses—all harmful or dangerous for humans to handle.
The main problem was designing a method to make electronic skin emulate human skin. Scientists are only beginning to understand how human ‘sensors’ such as mechanoreceptors and thermoreceptors work. In fact, the 2020 Nobel Prize in Medicine and Physiology was awarded to two scientists for discovering the neural mechanisms behind human sensations. What would be the technological equivalent of millions of nerve endings in human skin conferring incredible sensitivity and a wide range of detecting abilities, and how could such a system be built in a reproducible, scalable way?
With the help of colleagues with expertise in materials science and nano-engineering, Gao developed electronic skin, which he called E-skin-H. “My primary inspiration comes from human skin,” Gao said. Touch and pressure cause electrical changes that can then be converted to computer signals. To mimic the vast nerve network of human skin, Gao created sensor arrays with microscopic radii of detection, increasing both the sensitivity and strength of a signal compared to using a single large sensor. These properties are helpful in complex applications like detecting the presence of a specific chemical out of a large mixture.
Embedding such minuscule sensor arrays in a highly flexible matrix is a technically easier way to create a bridge between sensors and robots than integrating the existing large chemical sensors meant for analysis of dry particles on robots. The flexibility of the sensor array material enables E-skin-H to retain its sensing capabilities regardless of how the actual robotic hand or arm on which it is mounted moves. Moreover, by using a hydrogel underneath the e-skin interface, the robotic skin can test for chemicals like they are in solution even though the sensor array is technically in a solid state. For example, biochemical tests like ELISAs can detect specific proteins, like those marking the surfaces of a SARS-COV-2 virus, in a solution. But now, with the hydrogel, the sensor array can detect proteins from a solid surface. Perhaps the most brilliant facet of Gao’s e-skin is that it can be printed with an inkjet printer. It requires only a series of nano-material metal inks such as gold, silver, and platinum to decorate graphene electrodes and a 3D-hydrogel printer. The ease of production vastly increases the scalability, adaptability, and replaceability of the sensor arrays. It also decreases production costs, allowing for the creation of larger sensor arrays that can be modified to test for various new compounds.
In his article published in Science Robotics, Gao built a robot called M-bot, which used machine learning to learn how human muscles, specifically hand and arm muscles, move in response to detecting certain tactile or chemical threats. Gao evaluated E-skin-H to see how the skin’s sensing capabilities could assist robots in making AI-based decisions the same way a human would. He concluded that M-bot could be used to detect compounds in a contaminated environment and track the source of trace amounts of hazardous compounds. In an early demonstration, M-bot tracked a nerve agent leak in an open field by detecting a gradient of the compound across the sensor array. By tracking the location of the highest concentration of the compound, the AI algorithm within M-bot was able to signal to the motors within the robot arms and fingers to extend towards the location of the highest concentration to grasp objects and collect samples. “It was pretty impressive since it was fully automatic,” Gao said.
Gao sees E-skin-H being used in medical and defense applications within the next five to ten years. “You don’t want to send a human into a danger zone to detect explosives or biohazards. Electronic skin can be used in military, environmental, and agricultural applications. We just have to make the robots [that use e-skin] smarter and more automatic—with the help of materials science, chemistry, and data processing.” Gao said.
With the foreseeable future wrought by AI applications and robotic sensing technologies, Gao encourages anyone interested in robotics and technology to nurture this interest by identifying a problem, understanding where there is a gap in current technologies that attempt to address this problem, and imagining potential solutions. “Becoming involved in engineering or robotics is not a problem about technology, but more about developing a pattern of thinking. Trying competitions like FIRST Lego League and VEX Robotics inspires young students to imagine the fullest potential of robots,” Gao said. With this principle in mind, Gao combines chemical, medical, electrical, and mechanical principles to create solutions to the problems that fascinate him. As he expands his projects, from robots that use electronic skin in defense applications to nano-robots that deliver drugs to cells in the body, Gao is convinced of their increasing necessity in the future. “I’m excited to see how we interact with robots in our daily lives going forward,” Gao concluded.