How would you develop an intelligent robot



08.04.2021 11:53

The next generation of AI robots need intelligent materials

Dr. Michael Hagmann communication
Empa - Federal Materials Testing and Research Institute

Artificial intelligence (AI) is now seen across all industries as fundamental for the development of new types of products and technologies and gives companies that have mastered them a competitive advantage. What is still missing, however, is an in-depth discussion of the effects on society in order to better understand opportunities, but also possible risks. In an interview, the two Empa researchers Mirko Kovac and Aslan Miriyev from the Center for Robotics outline the direction in which AI and robotics research is developing.

You are about to organize a virtual international workshop entitled “Material Intelligence” - how can a material become intelligent? How do we have to imagine a material that you would describe as "intelligent"?

Mirko Kovac: The general definitions of “intelligence” typically refer to the interaction between an “agent” and its environment, e.g. the ability to learn, understand, react or adapt to situations and changes in the environment. An agent can be any organism, such as a fish, a cat, or a human. In our case, the agent is a robot. With regard to materials, intelligence lies in their ability to react to external stimuli. The challenge, however, is to turn this from a passive, more or less pre-defined response to stimuli to an intelligent adaptation to a changing environment. This means that, for example, a material that reacts to heat by changing its shape does so by adapting to the specific prevailing external conditions. We cannot change the laws of physics, but we can help the material through clever design to adapt to the situation at hand. So when we talk about material intelligence, we tend to think of a responsive system of matter and clever design.

Aslan Miriyev: An essential advantage here would be the reversibility, i.e. the reversibility, and a high repeatability of the intelligent behavior. Such systems could be suitable as the basis for a new type of computer architecture that uses shapes and material properties as a means of computing. In the workshop we discuss this path from functional materials as passive units that react to external stimuli to a system with which complex calculations can be carried out.

What is the connection between smart materials and artificial intelligence (AI)? And where or how would robotics fit into the picture?

Miriyev: In connection with the previous question, material systems form the physical artificial intelligence of robotic bodies, similar to the way biological tissue makes up the intelligence of our body. Synthetic and biohybrid materials are the key to collaborative robots that will coexist with humans in future symbiotic human-robot ecosystems.

In recent years, AI research has developed from an algorithm-based focus on the greatest possible computing power to a more physical approach to intelligence, the so-called physical AI. Where is this type of research headed?

Miriyev: Indeed, the field of AI has grown tremendously over the past few decades. It has become clear that AI is not just the digital AI that we know, for example, from the ability to use camera images to identify a pack of chips in a jumble of many objects on a table. AI is also about physical interaction with the world, with people, the environment, nature, plants, animals.

Kovac: The research is most likely going in the direction of developing the physical side of AI and merging it with its digital counterpart. We recently published these ideas in an article in the journal Nature Machine Intelligence. We define Physical AI (or PAI) as the theory and practice of developing physical systems capable of performing tasks typically associated with intelligent organisms.

What area or industry could benefit most from this type of AI? Do you have specific examples?

Kovac: The areas of application of physical AI include healthcare, elderly care, digitization of infrastructure, disaster management, public safety, the service sector, education and industrial automation. The entire Industry 5.0 concept is based on real human-robot interaction. So far, most of the successively commercialized soft robot solutions have been in the areas of pick-and-place, wearables, prostheses and minimally invasive medical applications.

Miriyev: The current use of Physical AI is still very limited compared to the challenges of the world around us and the opportunities that lie ahead. There are almost endless possibilities for the use of robots that are able to live and work together with humans.

You're working on autonomous drones, or robots, if you will, that are inspired by biological principles. How would you define their role, their output in the interaction with us humans?

Kovac: In 2013 I published an article entitled "The Bioinspiration Design Paradigm: A Perspective for Soft Robotics" (https://doi.org/10.1089/soro.2013.0004). The principle can be divided into three phases: inspiration, abstraction and implementation. In the inspiration phase, a scientist observes a natural phenomenon. In the subsequent abstraction phase, he or she tries to find out the underlying physical principles of the observed phenomenon. Then comes the implementation phase, in which the scientist builds the abstract bio-inspired design principles into a synthetic system. Specifically, we are inspired by the process of creating new living organisms as a new paradigm for creating artificial systems.

Miriyev: When building physical AI systems, knowledge from several disciplines helps create structures, sensors, actuators and the basis for calculating intelligent behavior at the same time. In the workshop we want to discuss different ways to such an advanced robot design through the use of different aspects of material intelligence.

In view of the many dystopian sci-fi films and novels, one has to expect a certain skepticism when speaking of the symbiosis of man and machine. What are the main challenges in this area and how do you approach them?

Kovac: For all our love of fictional dramas, we often forget that the demands of real life are much more obvious than the plots of science fiction scenarios. Older people actually need reliable help that goes well beyond telepresent humanoids.

Miriyev: Society needs lifelike robotic assistants who adapt in style, speed and language to both their older human companions and their surroundings in order to provide them with all the help they need. This should be done without endangering people's quality of life or their privacy and personal data. We have to concentrate on creating useful objects for people and for their everyday lives.

So where do we stand in, say, 20 years in terms of AI, robotics and the like?

Kovac: Robotics is becoming an integral part of our physical world and a digital fabric of our infrastructures and cities. Robotics as a term will disappear in many ways, and we will get used to dealing with artificial intelligence in our everyday lives. The challenge for the next few decades will lie in bio-hybrid systems, in the combination of robotics with biological materials. We are also working on this topic at Empa's Robotics Center.

Interview: MariaLucia Hijar, Empa Academy

Material Intelligence workshop
Attend the workshop "Material Intelligence: From Functional Materials to Material Computing" on April 12, 2021. You can find more information here: https://www.material-intelligence.com/


Scientific contact:

Prof. Dr. Mirko Kovac
Materials and Technology Center of Robotics
Tel. +41 58 765 4689
[email protected]

Dr. Aslan Miriyev
Materials and Technology Center of Robotics
Tel. +41 58 765 4631
[email protected]


Additional Information:

https://www.empa.ch/web/s604/materials-intelligence Empa homepage
https://www.material-intelligence.com/ Registration workshop


Features of this press release:
Journalists
society
supraregional
Scientific meetings
German