Internal Pages

Login to access

CoSwarm

From KOVAN

Image:Coswarm-logo.jpg

In the near future, advances in technology will pave the way for the mass production and deployment of large numbers of autonomous robots that can sense and physically interact with their environment. In order to operate swarms of robots for the surveillance of an environment or for the dynamical placement of communication nodes in an environment, effective coordination strategies have to be developed. It is shown that self-organization lies behind the successful coordination of natural distributed systems such as ant colonies or bee hives and that self-organization is accepted as a promising approach for the coordination robotic swarms. Most of the ongoing studies have focused on the application of self-organization to robotics swarms and the evaluation of sclability, robustness and performance achieved by the approach. The limitations of controllability of these swarms due to the use of the self-organization approach has been neglected so far, leaving the question of how useful the approach can be in real-world use, unanswered.

The main scientific objective of this project is to investigate how and to what extend the dynamics of a robotic swarm can be externally controlled. In the project, a heterogeneous swarm, consisting of two types of mobile robots, one in large numbers but simple, the other in fewer numbers but more complex, will be developed. The experiments to be made with the real robots will be complemented by systematic experiments carried out in physically realistic simulation models that will also be developed. In the first part of the project, the problem of controlled dispersion will be studied. This part will consist of two stages.In the first stage, the dipersion dynamics of the homogeneous swarm, made of only the simple robots, will be analyzed. Through systematic experimentations with the simulation model, the behavioral parameters of the robots will be varied and their effect on the swarm will be observed. The results obtained will be justified using the real robotic swarm. In the second stage, we will investigate how the dynamics of the simple robot swarm can be controlled throught the behaviors of the complex robot swarm mixed. The studies to be carried out can, in some aspects, resemble to the control of a large sheep flock by a group(swarm) of shepherd dogs. We will be seeking the coordination strategies for the shepherd dogs such that they can increase or decrease the dispersion density of the sheep flock, or push the sheep into a closed arena through a small gate. In the second part of the project the problem of scalable and fault-free self-organized pattern formation will be studied. Two ideas, inspired from natural systems will be investigated: a)~Similar to the formation of proteins from RNA in cells, we will explore how templates formed by the complex robots can guide the pattern formation process of simple robots. b)~Inspired from chemical reactions, we will explore the feasibility of using the complex robots as a ``catalyst for the simple robots to create a desired pattern. In both of these studies, the behaviors desired for the complex robots will be explored by a distributed evolutionary algorithm in the simulation environment, and the discovered behaviors will be verified on the real robotic swarms.

The project, with its interdisciplinary nature, is likely to create interest from different engineering disciplines and sciences increasing the intereaction among them. The distributed approach proposed in the project has emerged as an alternative problem solving approach to the traditional centralized approach. An indirect effect of the project is the promotion of this new approach. Some of the robots to be developed within the project will be used for the laboratory studies of an existing autonomous robotics course. Since the robots will be developed in partnership with a company, we will look into the possibility of licensing the design to the company to manufacture and sell it as cheap robotic kits to be used in robotic education at the universities (and possibly in high schools).

This page was last modified on 21 December 2009, at 15:47. This page has been accessed 3,912 times.