Home / Blog / Self-repairing robots do have a sense of self? – III

Self-repairing robots do have a sense of self? – III

System diagram for the bootstrap path of a self-replicating manufacturing system. Each subsystem in a blue line (as well as the entire system) meets the self-replicating system criterion. Image credit: J. Storrs Hall, Institute for Molecular Manufacturing, Los Altos, CA, US

I pointed out in the previous blog that a true self replication shall involve both top down and bottom up processes. Researchers are taking this system-wide approach in (nanotech) molecular manufacturing, as shown in the diagram. The industrial issue with nanotech and molecular manufacturing is that the yield of the manufacturing process would be infinitesimally low if you have to combine molecules by molecules into the desired structure. What is needed is an approach where once a desired molecule has been “manufactured” this molecule can start creating other molecules and so on generating an avalanche process that yields appreciable quantity of material.

There is another ongoing approach towards self-replication progressing through the intermediate steps of self-repair. Japanese researchers are working to create/teach robots to self repair (watch the clip at the end of this post). This is in certain way simpler, you are not aiming at self-duplicating an object, rather to endow that object the capability of self-fixing problems. This may require the creation of a component, the fixing of something (like tightening a screw) or the addition of a part. It requires something more: the awareness that something is needed!

This is what has brought me to put the question “Self-repairing robots do have a sense of self?”. In this respect, self replication can be a simpler process. Molecules, DNA, do not need to be aware of the need of replicating, In the end it is a mechanical process, their structure is inevitably leading, under certain condition, to replication. There is no need for awareness.

The situation for a robot is slightly different. A robot that is experiencing a difficulty needs to understand at a certain level what the problem is and then move the next step into finding a way to fix the problem. Clearly we can pre-program a number of situation in the robot to make this “awareness” nothing more than a coded response to a certain situation (holding an object is becoming difficult, clean the pinches to remove any dirt, if that is not sufficient tighten a screw on the pinch, ….). This kind of awareness is not really … awareness.

However, as the robot becomes more sophisticated and/or has to face an unpredictable context (like an autonomous robot designed to operate on Mars), then a higher level of awareness will become important.

The new wave of creating robots with self-teaching capabilities is implicitly creating robots that acquire a (although limited) sense of self. They have to realise the boundary/distinction between them and their environment and this implies the understanding of their own existence. As shown in the following clip, scientists at Columbia University have been able to create a robot that can self-learn its own capabilities and how to use them. This is a first step in creating a robot that can realise when something fails and need to be fixed.

Notice that awareness does not necessarily implies consciousness (I guess everybody can agree that a spider is aware but not all would agree a spider has a consciousness…). When we consider robots and work on making them ever more autonomous, something that is an absolute requirement for space exploration but also for deep ocean exploration by rovers, we see that engineers are working to improve as much as possible their context awareness and to embed a sense of self to the extent that they can evaluate their own functionality with respect to the context and the goal and take autonomous decisions. Is this leading to consciousness? The opinions are quite broad on the subject and have been discussed over and over in the area of computer consciousness (or lack of it).

Progress in autonomous systems, in robotics and in their application space are fuelling discussion and I guess we are going to see more of them in the next decade. Whether these discussions will lead to an accepted answer remains to be seen.

The answer to my original question, however, may be more straightforward: Yes! In order be be able to self repair  in a broad sense robots need to have a sense of self. This sense may not be needed when the repair is minimal and mechanically controlled/implemented but it will have to grow as repair becomes more complex. Our own body has plenty of self-repairing capability and most of them occur without ourselves being conscious. In some cases the repair requires some consciousness, like noticing your finger is bleeding and using a patch (or a stitch) to fix it. Some other time more advanced (intellectual and consciousness) capabilities are required, like when a climber plans for a high mountain climb. She understands the need for increasing the red cells in her blood stream and starts an acclimatization process. This implies knowledge of the challenge ahead, understanding of risks, evaluation of possible countermeasure and execution of specific steps over a long period of time. This is clearly a process implying a high level of awareness. Something similar will be required by robots.

About Roberto Saracco

Roberto Saracco fell in love with technology and its implications long time ago. His background is in math and computer science. Until April 2017 he led the EIT Digital Italian Node and then was head of the Industrial Doctoral School of EIT Digital up to September 2018. Previously, up to December 2011 he was the Director of the Telecom Italia Future Centre in Venice, looking at the interplay of technology evolution, economics and society. At the turn of the century he led a World Bank-Infodev project to stimulate entrepreneurship in Latin America. He is a senior member of IEEE where he leads the Industry Advisory Board within the Future Directions Committee and co-chairs the Digital Reality Initiative. He teaches a Master course on Technology Forecasting and Market impact at the University of Trento. He has published over 100 papers in journals and magazines and 14 books.