Corresponding author:

Ronzhin Andrew L., D. Sc., Professor, Deputy Director, St. Petersburg Institute for Informatics and Automation of the Russian Academy of Sciences, St. Petersburg, 199178, Russian Federation. E-mail: ronzhin@iias.spb.su

Received on August 12, 2016
Accepted on September 01, 2016

The considered problems of aggregation of the swarm robots are mainly connected with the simplest computing, sensors and built-in actuators, as well as limited resources of the homogeneous swarm robots. In the area of the swarm robotics the multi-agent technologies are used to simulate the interaction of big groups of simple homogeneous robots. The limited resources of the individual robots have a significant effect on the configuration and capabilities of the whole system, however, the distributed swarm intelligence based on the data obtained during the mass pair interactions of the robots ensures the existence of a swarm and, due to it, solving of the set tasks. A review of the main methods for solving of the problem of aggregation in a swarm of robots (the method of the virtual forces, the probabilistic and evolutionary methods) showed that the choice of the used method, first of all, depends on the sensor, computing and network resources of the robots. Examples of different aggregation algorithms for the robot swarms, as well as limitations on the design solutions during implementation of these algorithms are presented.
For estimation of the effectiveness of the robot aggregation the spatial and temporal methods are mainly used. The specific choice of the metrics depends on the parameters of the form, which should be achieved, as well as the aggregation method which is technically possible to implement to ensure control of a given robot swarm. By the results of the analysis, a conclusion was made that the systems, which use the neural networks for the swarm control are the most promising in terms of the further improvement of the aggregation algorithms, however, their implementation requires large onboard computational resources. During the preliminary research concerning the swarm robotics the principles and conceptual model were developed for the process of reconfiguration of the spatial position of the group of robots. They take into account the restrictions on the geometrical dimensions of a set of homogeneous robots and the occupied area of the initial position of the robots, the spatial characteristics, density of the robots' locations, as well as the ways of defining of the target position coordinates of the robots in a new spatial configuration. The further research will be aimed to solve the tasks of interaction within the robot swarm during construction of more complex forms, taking into account a bigger number of the physical parameters.
Keywords: swarm robotics, robot swarm, aggregation, aggregate, decentralized control, multi-agent technology, reconfiguration, embedded board computers, multimodal sensors

Acknowledgements: The study was supported by RFBR (grant № 16-29-04101-ofi_m) and the draft program of the Presidium of the Russian Academy of Sciences I.31P "Actual Problems of Robotics."

For citation:

Shlyakhov N. E., Vatamaniuk I. V., Ronzhin A. L. Review of the Methods and Algorithms of a Robot Swarm Aggregation, Mekhatronika, Avtomatizatsiya, Upravlenie, 2017, vol. 18, no. 1, pp. 22—29.
DOI: 10.17587/mau.18.22-29

To the contents