Swarm Intelligence for Space Applications
What is swarm intelligence?
Actually it is difficult to give a definition - the scientific community does not agree on one yet. We at the ACT define swarm intelligence as the emerging property of systems made by multiple identical non-cognitive agents characterized by limited sensing capabilities. In other words: When a set of simple machines perform simple behaviours we observe that the outcome may not be at all simple. Indeed, this effect of emerging behaviour resembles that of biological swarms, where a number of rather un-intelligent animals may form a swarm and hence achieve more than the sum of the individuals could do. Swarm intelligence, as a research field, has come from the observation of swarming animals and from the attempts of the scientists to mimic their behaviour. As an example of its biological counterpart, think of some types of fish that show a certain bell shaped swarm configuration when they feel attacked in order to frighten and keep away potential predators.
How can we do it?
As turning off the gravity field is impossible, only some tricks allow to simulate weightlessness without going to space. As submerging under water is a bit difficult with an electrical setup, we decided to go for the free-falling approach. The safest way to do so, is to go on an aeroplane that performs flight maneuvers which have a free-falling component. The Airbus A300 0G-aircraft from Novespace in Bordeaux is the world's largest one for parabolic flights. It has been used since 1997 by ESA, CNES, DLR and industrial customers to provide repeated microgravity periods of up to 20 seconds for research purposes.
Why in space?
A number of features of swarm intelligence are attractive for the space engineering community. The space environment typically puts stringent constraints on the capabilities of single satellites, robots or anything that needs to survive in space. Space agents are particularly limited in terms of mobility (propellant and power limited), communication (power limited) and size (mass limited). At the same time, a high level of adaptability, robustness and autonomy is required to increase the chances of success of operating in a largely unknown environment. Similar characteristics are found in the individual components of a biological swarm.
Furthermore, a number of space applications are based on the presence of multiple space agents. The very first commercial application proposed and realized for satellite systems was that of Arthur C. Clarke and was a satellite constellation providing global communication services by means of three satellites put in a geostationary orbit. Since then, a large number of constellations have been deployed to provide global communication, navigation and Earth observation services. More recently, the idea of satellites flying in formation has also been used in a number of missions for applications ranging from X-Ray astronomy, differential measurements of the geomagnetic field, space interferometry, to the search for exoplanets. Swarm intelligence methods represent an attractive design option allowing autonomous operations of multiple spacecraft be achieved. Simpler agents with limited capabilities could be considered as a resource, rather than as an overhead. At the same time, one is able to engineer systems that are robust, autonomous, adaptable, distributed and redundant. In addition, swarms allow for mass production of single components and represent highly stowable systems.
What applications is the ACT looking at?
As an orbital swarm has a collective consciousness coded in its control system, we may imagine, in the long term, that it may be possible to send hundreds of small satellites into orbit to build a large antenna just as birds build their nests. In a closer future, swarm intelligence techniques can help us to achieve decentralized control systems for coordinated planetary exploration or spacecraft formation flying.
The ACT is involved in a number of research activities in swarm intelligence which are described on the Artificial Intelligence pages of this website. At the moment we are investigating, together with the University of Bruxelles, how to evolve the control systems of the agents based on the collective achievements of the swarm. The results of this study, part of the Ariadna scheme, will be ready in early 2008.
An ACT Podcast on the subject of swarm intelligence for space application is also available for download, try it!!!