New research suggests combining educational subjects and research disciplines to help researchers breathe life into intelligent robots made.
THE comment suggests that the science of teaching materials, mechanical engineering, IT, biology and chemistry as a combined discipline could help students develop the skills they need to create artificially intelligent robots (AI) as Researchers.
Physical AI is the new border of robotics research and will have a major impact in the decades to come. Professor Mirko Kovac Department of Aeronautics
Known as physical AI, these robots would be designed to resemble and behave like humans or other animals while having intellectual capacities normally associated with biological organisms. These robots could in the future help humans in work and in daily life, to carry out dangerous tasks for humans and to help medicine, care, security, construction and industry.
Although machines and biological beings exist separately, the intelligence capacities of the two have not yet been combined. Until now, there has been no autonomous robot that interacts with the surrounding environment and with humans in the same way as the current fact of the computer and the AI based on smartphones.
Co-directed author Professor Mirko Kovac imperial Department of Aeronautics and the Swiss federal laboratories for materials science and technology (Empa) Robotics material and technology center said: “the development of” body “of robots has considerably lagged behind the development of” brains “of robots. Unlike digital AI, which has been intensively explored in recent decades, their physical intelligence that has remained there relatively unexplored. »»
The researchers say that the reason for this gap could be that no systematic educational approach has yet been developed to teach students and researchers to create bodies of robots and whole combined brains.
This new research, which is published in Intelligence of the nature machine, Defines the term physical for the first time. He also suggests an approach to overcome the skills gap by integrating scientific disciplines in order to help future researchers create realistic robots with capacity associated with intelligent organizations, such as the development of bodily control, autonomy and detection at the same time.
Researchers will need a much wider stock of skills to create realistic robots, and interdisciplinary collaborations and partnerships will be very important. Professor Mirko Kovac Department of Aeronautics
The authors have identified five main disciplines which are essential to create a physical AI: materials science, mechanical engineering, IT, biology and chemistry.
Professor Kovac said: “The concept of AI is often confined to computers, smartphones and data data calculation. We propose to think of AI in a broader sense and to co-develop physical morphologies, learning systems, integrated sensors, fluid logic and integrated activation.
“Physical AI is the new border of research in robotics and will have a major impact in the decades to come. The skills of co-evolved students in an integrative and multidisciplinary way could unlock key ideas for students and researchers . “
The researchers say that the realization of natural type functionality in robots requires combining conventional robotics and AI with other disciplines to create a physical AI like its own discipline.
Professor Kovac said: “We are considering robots of advanced and cultivated physical AI using a variety of unconventional materials and research methods. Researchers will need a much wider stock of skills to build realistic robots, and interdisciplinary collaborations and partnerships will be very important. »»
An example of such a partnership is the Imperial-Empa articulation Robotics material and technology center which connects the expertise in material science of the EMPA with the Imperial Air robotics laboratory.
The development of skills and research in physical AI could bring us closer to redefining the interaction of human Obroben and the robot environment. Dr Aslan Miriyev Empa / Department of Aeronautics
The authors also offer the intensification of research activities in physical AI by supporting teachers at the institutional and community level. They suggest hiring and supporting teachers whose priority will be physical multidisciplinary research.
Co-directed author Dr Aslan Miriyev Of EMPA and the Imperial Aeronautical Department said: “Such support is particularly necessary, because working in the multidisciplinary playground requires leaving the areas of close disciplinary knowledge for high -risk research and a career uncertainty.
“The creation of realistic robots has so far been an impossible task, but this could be made possible by including a physical AI in the higher education system. The development of skills and research in physical AI could bring us closer to redefining the interaction of human Obroben and the robot environment. “”
The researchers hope that their work will encourage an active discussion on the subject and will lead to the integration of the physical disciplines of the AI in the current of education.
Researchers intend to implement the physical methodology of AI in their research and education activities to open the way for human robot ecosystems.
“Skills for physical artificial intelligence” by Aslan Miriyev and Mirko Kovac, published on November 10, 2020 Nature machine intelligence.
Image credits: Shutterstock (hand); Imperial College London / Empa