The Evolution of Cooperative Behaviours in Physically Heterogeneous Multi-Robot Systems

Jianhua Yang,Zhaohui Wu,Min Yao,Yabo Liu

doi:10.5772/53089

Abstract

Evolutionary Robotics (ER) is a methodology that uses evolutionary computation to develop controllers for autonomous robots. The existing works focus on single robot systems or physically homogenous multi-robot teams, although physical heterogeneousness is more prevalent in the real world. It is therefore instructive to examine whether cooperative behaviours can be synthesized using artificial evolution for a team of physically heterogeneous robots. This paper makes an important contribution in answering the question of whether robots with distinct capabilities can synthesize their control strategies to accommodate their own capabilities without human intervention. We present an empirical analysis of the collaboration mechanisms and suggest guidelines about how to choose appropriate evolution methods. Simulated experiments with a team of e-puck robots show that evolution can lead to effective controllers for robots with distinct capabilities.

Highlights

The unknown and dynamic characteristics of environments may mean that explicitly designing a control system for a single robot is difficult, let alone for multi‐robot systems (MRS)
The robots can perceive the intensity of sound using three sound sensors that simulate three directional microphones using a set of equations [9], and the total perceived amplitude (TPA), i.e., the result of the contribution of different sound sources corresponding to different robots is computed
The EP1 robots have greater speed and longer vision sensor ranges than the EP2, but their task solving capabilities are quite poor compared to the EP2

Summary

Introduction

The unknown and dynamic characteristics of environments may mean that explicitly designing a control system for a single robot is difficult, let alone for multi‐robot systems (MRS). Evolutionary robotics (ER) is a field of research that applies artificial evolution to generate control systems for autonomous robots [1, 2]. It is possible to learn and synthesize the control systems of a group of robots that are able to display expected behaviours without human supervision. A group of robots are evolved to display aggregation and coordinated movements [9,10,11], perform multi‐objective tasks [12, 13], and learn to cooperate to play competitive games [14,15,16,17], etc.

Methods

Results

Conclusion