Adaptation in natural and artificial systems
Adaptation in natural and artificial systems
Robot Motion Planning
Fairing of parametric cubic splines
Mathematical and Computer Modelling: An International Journal
Voronoi-initializated island models for solving real-coded deceptive problems
Proceedings of the 10th annual conference on Genetic and evolutionary computation
Hi-index | 0.00 |
Path planning and obstacle avoidancemethods are often required for robots working in more and more complicated environments. This paper introduces a novel approach called Voronoi Strains for solving this task. The algorithm applies particle swarm optimization of cubic splines which are connected to strings. The initialization of the evolutionary algorithmis based on the Voronoi graph method as well as on strains of bacteria. Strains whose evolution is stuck in a dead end because of a local optimum die off. Only the strain located close to the global optimum or the biggest local optimum will survive the evolutionary process. Different settings of PSO parameters have been tested in various simulation experiments. The Voronoi Strains approach was also compared with other PSO methods using a different kind of initialization.