Physical Artificial Intelligence

The concept of Physical Artificial Intelligence (PAI) proposed by Prof. Mirko Kovac from Aeronautics Department at Imperial College London (and a member of our advisory board), could redefine the field of robotics and the relationship between man and machine. Artificial Intelligence makes machines perform at ever more amazing levels. A growing group of researchers believes that the current “intelligence” is limited. It is too attached to data-driven trends in deep learning and what humans think of as intelligence, which often reflects loose thinking about human cognitive capacities. A robot that can do little more than a remote-controlled model car has a limited range of applications. But from an automatic machine to an autonomous robot, it is a big, almost revolutionary step. The combined discipline of PAI could effectively be the missing link in the attempt to create artificially intelligent robots that look and behave like humans. Only when the artificial intelligence of a digital "brain" is merged with an intelligent body could new types of robots be created. PAI takes inspiration from the complex behaviours and capabilities of biological organisms. The PAI approach offers the opportunity to incorporate homeostasis, which is seen as an important process for organisms to regulate their behaviour and adapt to different environments. In 2019, Man and Damasio from the University of South California observed that the field of soft robotics has advanced to a stage, where a process resembling homeostasis could be integrated with intelligent machines. This integration of a body, internal regulatory mechanisms and control could lead to a new class of machines that have intrinsic goals. PAI comprises five main disciplines: mechanical engineering, computer science, biology, chemistry and materials science. We can expect that what is generally considered as intelligence and AI will remain in flux. By integrating advances from different disciplines, there is an opportunity to create intelligent machines with ever greater complexity. PAI could make our life much easier. Quoted from Forbes, “If developed, lifelike autonomous robots could potentially help humans at work and in daily living, performing tasks that would otherwise be dangerous, onerous or tedious. They could, for example, assist in health care and social care, helping to plug the global shortage of healthcare workers.”

Source: Nature Machine Learning, 11/2020