The analysis of various methods of solving the navigation problem by an intelligent agent in the environment of virtual football Robosoccer on the basis of noisy data coming from a visual sensor. Two groups of methods for calculating the absolute coordinates of objects based on sensory data about flags and lines are considered: trigonometric methods and methods based on the use of the Kalman filter and the particle filter. The instrumental software tool developed for carrying out experimental researches and allowing to vary arbitrarily conditions of the decision of a navigation problem is briefly described. Experimental results of the comparative analysis of the speed and accuracy of algorithms implementing various methods are presented. The obtained results allow the agent to solve the navigation problem using algorithms of arbitrary time, exchanging the solution time for the quality (accuracy) of the result. Taking into account the obtained results, the directions of further research on the implementation of the assessment of the tactical situation in virtual football are determined.

Authors: M. G. Panteleyev, A. F. Salimov

Direction: Informatics, Computer Technologies And Control

Keywords: Intelligent agent, multi-agent system, virtual soccer, RoboCup Soccer, navigation task, arbitrary time algorithms, Kalman filter, particle filter

View full article