Evolucionārās skaitļošanas pielietošanas analīze mobila robota vadībai
NOTE: Thesis is in Latvian.
Thesis provides a practical overview of genetic algorithm use for neural network training with purpose to gain automatized robot control mechanism.
Theoretical part includes summary of theoretical basis for evolutionary computation and its applications in robotics.
Practical part includes created mobile robot imitation model and software for neural network creation by means of genetic algorithm.
Created tools were used to analyse different neural network topology applications for obstacle avoidance task. Conclusion contains discussion of attained results and identified topics for future research in this field. Appendix contains user’s guide for developed software.
This repository includes simulated robot and maze for it created in anyKode Marilou 2010.
The controller application is written in C# (Visual Studio project).