Living beings have shown an incredible capacity to adapt to their environment. That went through million of years, thousands and thousands of generations and immense extintions (see graph). The ones surviving are the ones that manage to adapt to a changing environment. Autonomous systems have been designed to operate within specific boundaries. As they become more and more powerful they will face broader and broader contexts and will need to adapt to ever more complex changes.
In turns, this requires a growing “understanding” of their environment (bordering on awareness) and the capability to alter the “rules of the game” through which they conform their behavior. In symbiotic systems this issues is compounded by the presence of two, or more, interactive autonomous systems, one of which can be a biological one. A further demand for adaptability derives from a possible failure in one or more components of the system. This failure should be managed by the system possibly with a degradation of functionality but preserving the overall capabilities. In symbiotic systems there is no “overall” system from the point of view of “reliability” Each of the autonomous systems composing the symbiotic entity is reacting to a failure of the other as a change of context and has to take countermeasures fitting its own goal, not the ones of the overall system. We have examples of this situation in living beings where the symbiotic relationship between ourselves and the bacteria ecosystem in our guts is essential to our wellbeing and there is very little latitude for an adaptation if the bacterial system fails.
This is an issue that needs to be faced in the design of symbiotic autonomous systems, for those part that are under our design control. We usually can design just one or a few components but not all of them. Hence there is the need of understanding what is the range of adaptation of those systems that are not under the designer control to make the best out of what can be controlled.
Notice that adaptation leads to evolution. A system that has adapted to a new contex will respond differently from the system “it used to be”. Over time adaptation pressure leads to the creation of quite different systems. In the future we will be seing systems that will be able to create new systems, new offsprings, that eventually will take over. We are moving the first steps in this direction with robots that can build better robots to face newer tasks as well as software that can better perform in the solution of problems.
There are already many areas where there is a need to design self-adaptive autonomous systems like in smart cities. Vehicular traffic flow can be seen as an autonomous system of its own, interacting to pressures that are coming from events attracting people in a certain area. People, as a whole, are themselves an autonomous systems and conditioning them to use one form of transportation or distributing them on several forms of transportation is an overall smart city design issue. These two systems interact one another and they also are directed by a context and by constrains and resource availability. The distribution of shopping malls, parking areas, the coordination of sales campaigns has an impact on these systems. The logistics (both supply and delivery) is also impacting the behavior of these systems.
A smart city needs to have a monitoring system that visualizes the various autonomous systems composing it and, this is the crucial point for the future, will have to orchestrate them to become a symbiotic system. By 2050 we can expect many ambients, like smart cities, to have become symbiotic systems, eco-biomes.