A Step-by-Step Algorithm for Finding the Optimal Strategy for the Behavior of a Group of Robots

Abstract

The solution of the multi-criteria problem, which includes the distribution of objectives, the planning of trajectories and the optimization of energy consumption, is considered in the realization of the collective interaction of robots. It is proposed to use a genetic algorithm according to the chosen conditions (constraints) and optimality criteria to find the best strategy for group behavior. The considerable difficulty in choosing how to control a team of autonomous mobile robots represents the distribution of tasks among agents that operate under conditions of parametric and information uncertainty, possess modest hardware, power and functionality. Therefore, the implementation of a multi-stage search for an optimal solution requires a specialized approach that takes into account the whole range of dynamic parameters, allowing for real-time target correction and degradation of robots until they fail. The basis of the proposed neurogenetic algorithm is a new algorithm for calculating the fitness function, in which the results of the neural network method of trajectory planning for a group of robots are used, as well as information about the initial charge of the batteries of robotic agents of the collective, the energy consumption of each agent and the preliminary estimation of the energy required by some agent to perform the individual tasks available to it. In order to ensure an acceptable performance of the algorithm and given the high dynamism of the external environment, it was decided to limit the search for solutions to only one step (the next working beat of the collective). The paper presents the results of the simulation of the task of finding the optimal behavior of robots, the algorithm of calculation of the specialized fitness function and the options of step-by-step search of the global strategy of distribution of tasks, which make it possible to increase the efficiency of the use of the team of robots due to the guaranteed production of the result while minimizing the total time of completion of all the tasks, as well as to increase the working time of the team due to the correct energy consumption.