Information Technology Reference
In-Depth Information
1. We showed how to adapt OpenStreetMap maps such that they can be used
for dynamic routing via ant colonies.
2. We showed that the published versions of ant algorithms tend to prefer routes
with nodes of lower degrees and developed a concept how to avoid that.
3. We showed that the statistically fuzzy behaviour of ant algorithms can be
used for trac management with the goal of equally distributing the trac.
This makes congestions less likely to occur.
Next work to be done towards an integration into real navigation systems is
the following:
1. The number of nodes extracted from OpenStreetMap can be further reduced:
Quite a few OpenStreetMap nodes physically belong to the same junction.
These nodes should be summarized to a supernode that serves as single
source or target for the ants of the colony.
2. In order to enable a higher degree of parallelism, the ant simulations should
be distributed to several computers. AntScout is perfectly prepared to this
using the actor concept of the Akka library.
3. In order to improve the requirements of real navigation systems, AntScout
should integrate other criteria than the mere passing time of road segments.
4. First attempts have already been published how to integrate hierarchies
into ant algorithms conceptually (cf. [10], [5]). These suggestions should be
elaborated and integrated into AntScout.
While the first three items are matters of implementation which do not require
conceptual adjustments, the last item requires more conceptual research.
The physical visibility of the functionality of AntScout shows that a concep-
tually innovative and also very exotic paradigm such as ant colonies may be
used for real world applications. We are convinced that navigation systems of
the future will apply the following advantages of ant algorithms (elaborated in
Section 2.2) regardless if they are called ant-based or not:
1. eager computing
2. off-board middleware
3. statistical compression of individual information
4. local information first
References
1. Bertram, A.: Ant Scout - Dynamisches Routing auf OpenStreetMap. Master's
thesis, FH Wedel (2012) (in German),
http://www.fh-wedel.de/fileadmin/
2. Blocker, C., Iwanowski, S.: Utilizing an ant system for a competitive real-life plan-
ning scenario. In: COMPUTATION TOOLS 2012: Third International Conference
on Computational Logics, Algebras, Programming, and Benchmarking, pp. 7-13
(2012)
