Evolving flocking in embodied agents based on local and global application of Reynolds’ rules

Abstract

In large scale systems of embodied agents, such as robot swarms, the ability to flock is essential in many tasks. However, the conditions necessary to artificially evolve self-organised flocking behaviours remain unknown. In this paper, we study and demonstrate how evolutionary techniques can be used to synthesise flocking behaviours, in particular, how fitness functions should be designed to evolve high-performing controllers. We start by considering Reynolds’ seminal work on flocking, the boids model, and design three components of a fitness function that are directly based on his three local rules to enforce local separation, cohesion and alignment. Results show that embedding Reynolds’ rules in the fitness function can lead to the successful evolution of flocking behaviours. However, only local, fragmented flocking behaviours tend to evolve when fitness scores are based on the individuals’ conformity to Reynolds’ rules. We therefore modify the components of the fitness function so that they consider the entire group of agents simultaneously, and find that the resulting behaviours lead to global flocking. Furthermore, the results show that alignment need not be explicitly rewarded to successfully evolve flocking. Our study thus represents a significant step towards the use of evolutionary techniques to synthesise collective behaviours for tasks in which embodied agents need to move as a single, cohesive group.

A unifying method-based classification of robot swarm spatial self-organisation behaviours

“…The first is to arbitrarily choose the different weights to obtain the desired behaviour. Another solution is to use optimisation techniques such as reinforcement learning (Hahn et al, 2019) or artificial evolution (Wood & Ackland, 2007; Vásárhelyi et al, 2018; Ramos et al, 2019) to obtain the optimal weights that best meet the desired optimisation function. The last solution allows the weights of the rules to be adapted during the simulation.…”

“…Attraction, alignment and repulsion (Reynolds, 1987) ( Reynolds, 1987;Fetecau, 2011;Yasuda et al, 2014;Cheraghi et al, 2020;Couzin et al, 2002;Aoki, 1982;Wilensky, 1998;Moeslinger et al, 2009;Leonard & Fiorelli, 2001;Fierro et al, 2001;Ferrante et al, 2012;Bonnefond et al, 2021;Stranieri et al, 2011;Hahn et al, 2019;Wood & Ackland, 2007;Vásárhelyi et al, 2018;Ramos et al, 2019;Hoang et al, 2021) (Mathews et al, 2012) (Couzin et al, 2002) Preservation of connectivity (Khaldi et al, 2018(Khaldi et al, , 2020Berlinger et al, 2021) (Tanner et al, 2003a(Tanner et al, , 2003b(Tanner et al, , 2007Olfati-Saber, 2006;Zavlanos et al, 2007;McCook & Esposito, 2007;Su et al, 2009;Wen et al, 2012;Ning et al, 2017;Dai et al, 2019;Gu & Hu, 2008;Yu et al, 2010) (McLurkin & Smith, 2004De Silva et al, 2005;Ugur et al, 2007;…”

Hierarchical Control of Smart Particle Swarms

“…In some applications, robots are required to move while maintaining a pattern or formation. This behavior is usually referred to as flocking [18], [19], taking inspiration from the behavior of birds. Flocking is extensively studied in literature, with hundreds of different approaches [20] inspired from the basic microscopic model [21] of attraction, repulsion and cohesion.…”

The effects of time-delay and phase lags on symmetric circular formations of mobile agents

“…The problem of controlling groups of mobile agents has several applications in agriculture [1], data collection [2], and space engineering [3,4]. Implementations range from agent-based systems [5,6] and artificial neural networks [7,8] to coupled-oscillator dynamics [9,10].…”