Scientists from Nanyang Technological University, Singapore (NTU Singapore) have developed, using a brain-inspired approach, a way for robots to recognize pain and self-repair when damaged by artificial intelligence ( AI).
The system has AI-enabled sensor nodes for processing and responding to ‘pain’ arising from physical force pressure. The system also enables the robot, without the need for human intervention, to detect and repair its own damage when minorly ‘injured’.
Currently, robots use a network of sensors to generate information about their immediate environment. For example, a disaster rescue robot uses camera and microphone sensors to locate a survivor under debris and then pulls the person out with guidance from touch sensors on their arms. A factory robot working on an assembly line uses vision to guide its arm to the right location and touch sensors to determine if the object is slipping when picked up.
Typically, today’s sensors do not process data but send it to a single large, powerful, central processing unit where learning happens. As a result , current robots are typically heavily wired, resulting in delayed reaction times. They are often vulnerable to damage that can be lengthy and expensive, requiring maintenance and repair.
The new NTU approach integrates AI into the sensor node network, linked to several small, less-powerful, processing units that act as ‘mini-brains’ distributed on the robotic skin. This means that learning takes place locally and, compared to traditional robots, the wiring requirements and response time for the robot are reduced five to ten times, say the scientists.
Combining the device with a form of material for self-healing ion gel ensures that the robots can restore their mechanical functions without human intervention when damaged.
The NTU scientists’ groundbreaking study was published in August in the peer-reviewed scientific journal Nature Communications.
“Co-lead author of the research, Associate Professor Arindam Basu of the School of Electrical & Electronic Engineering said, “One problem for robots to work with humans one day is how to ensure that they communicate with us safely. Scientists around the world have therefore found ways to offer robots a sense of awareness, such as being able to ‘feel’ pain, respond to it, and react to it.
According to Assoc Prof Basu, a neuromorphic computing expert, “Our work has shown the feasibility of a robotic system capable of processing information efficiently with minimal wiring and circuits. Our system should become affordable and scalable by reducing the number of electronic components required. This will help speed up the adoption of a new generation of robots in the future.”
The robust system allows the ‘injured’ robot to repair itself.
The research team created memtransistors, which are ‘brain-like’ electronic devices capable of memory and data processing, as artificial pain receptors and synapses to teach the robot how to perceive pain and learn harmful stimuli.
The research team showed through lab tests how the robot was able to learn in real time to adapt to injury. They also showed that even after injury, the robot continued to respond to pressure, showing the system’s robustness.
The robot easily loses mechanical control when ‘injured’ with a cut from a sharp point. But the molecules begin to interact in the self-healing ion gel, allowing the robot to ‘stitch’ its ‘wound’ together and restore its function while retaining high reactivity.
Rohit Abraham John, who is also a Research Fellow at the School of Materials Science & Engineering at NTU, first author of the study, said, “The self-healing properties of these new devices help the robotic system to repeatedly stitch together even at room temperature when ‘injured’ with a cut or scratch. This mimics how our biological system works, just like how human skin heals on.”
Our robot will ‘survive’ in our tests and respond to accidental mechanical damage resulting from minor injuries such as scratches and bumps while continuing to operate effectively. If such a device is used in real world settings with robots, it might lead to maintenance savings.
“Associate Professor Nripan Mathews, who is co-lead author and from the NTU School of Materials Science & Engineering, said,” Conventional robots perform tasks in a standardized programmable manner, but we can perceive their environment, learn and adjust actions accordingly. Most scientists concentrate on creating more and more responsive sensors, but do not concentrate on the challenges of how they perceive their environment, learning and behavior.
Our team has taken an off-the-beaten path approach in this work, applying new learning materials, devices and manufacturing techniques for robots to imitate human neuro-biological functions. Our findings have laid down important frameworks for the field, while still at a prototype stage, pointing the way forward for researchers to address these challenges.
Building on their previous body of work on neuromorphic electronics such as using light-activated devices to recognise objects, the NTU research team is now looking to collaborate with industry partners and government research labs to enhance their system for larger scale application.
Source of Story: Nanyang Technological University provides materials. Note: For style and length, material can be edited.
No CommentsLeave a comment Cancel