240
■
Bio-Inspired Computing and Networking
based only on local information, and feed the estimated values into the framework
described above.
9.6 Conclusion
We reviewed some basic ideas and preliminary results allowing for fully automated
deployment of arbitrarily large swarms from high-level and rich specifications. Our
approach is hierarchical. In the first level of the hierarchy, we aggregate the large
dimensional state space of the swarm into a small dimensional continuous abstract
space which captures essential features of the swarm. In the second level, we control
the continuous abstraction so that specifications given in linear temporal logic
over linear predicates in the essential features are satisfied. For planar robots with
polyhedral control