Optimisation in Dynamic Environments

Opimisation methods search for an optimal solution with regard to a target function. Among other properties, the target function is typically assumed to be static in the sense that the values of the same argument but calculated at different times do not change systematically. Yet many interesting real-world problems are not static but may change either subtly or substantially over time. Examples are real time scheduling problems were new jobs arrive with high frequency, real-time traffic planning, hardware settings with systematic movements and mechanical wear-off, or natural influences such as changing temperature or wheather conditions.

Having EvA2 as a powerful framework for heuristic optimisation, we are interested in developing and testing methods which handle such dynamically changing optimisation problems using evolutionary and other biologically motivated strategies. The PSO method, which employs swarm intelligence concepts for optimisation, is especially promising in this domain and constitutes a central theme within the project.

Comparison of several optimisers on a quickly changing benchmark function, by Geraldine Hopf

References

  • T. Blackwell, J. Branke. Multi-Swarms, exclusion, and anti-convergence in dynamic environments. IEEE Transactions on Evolutionary Computation 10 (4): 459-472. 2006.
  • R. Morrison. Performance measurement in dynamic environments. In J. Branke, editor, GECCO Workshop on Evolutionary Algorithms for Dynamic Optimization Problems, pages 5-8, 2003.

Publications

  • M. Kronfeld, C. Weiss and A. Zell: Swarm-supported Outdoor Localization with Sparse Visual Data. Proceedings of the 3rd European Conference on Mobile Robots (ECMR 2007), Freiburg, Germany, September 19-21, 2007, pp. 259-264.

Contact

Marcel Kronfeld, Tel.: (07071) 29-78987, marcel.kronfeld at uni-tuebingen.de