Evolution of Collective Behaviors for a Real Swarm of Aquatic Surface Robots

Abstract

Swarm robotics is a promising approach for the coordination of large numbers of robots. While previous studies have shown that evolutionary robotics techniques can be applied to obtain robust and efficient self-organized behaviors for robot swarms, most studies have been conducted in simulation, and the few that have been conducted on real robots have been confined to laboratory environments. In this paper, we demonstrate for the first time a swarm robotics system with evolved control successfully operating in a real and uncontrolled environment. We evolve neural network-based controllers in simulation for canonical swarm robotics tasks, namely homing, dispersion, clustering, and monitoring. We then assess the performance of the controllers on a real swarm of up to ten aquatic surface robots. Our results show that the evolved controllers transfer successfully to real robots and achieve a performance similar to the performance obtained in simulation. We validate that the evolved controllers display key properties of swarm intelligence-based control, namely scalability, flexibility, and robustness on the real swarm. We conclude with a proof-of-concept experiment in which the swarm performs a complete environmental monitoring task by combining multiple evolved controllers.

Controle baseado em redes neurais artificiais para agentes móveis em formação

“…Por outro lado, a ideia aquié que a rede neural aprenda as interações por meio de técnicas evolutivas. Para geração da rede neural, foi utilizada uma função objetivo, descrita pela Equação 2, que leva em conta a distância entre os agentes e o AV, e também um termo S que penaliza a função na iminência de uma ou mais colisões, no decorrer da simulação [3]. O sentido de maximização da função depende de os agentes chegarem o mais próximo possível do AV, e evitarem colisões entre si.…”

“…Li e Liu [6] abordaram o problema de controle de robôs quadrúpedes, e Habibi et al [5] trabalharam com o problema de transporte de um objeto por múltiplos robôs trabalhando coletivamente. Duarte et al [3] desenvolveram controles para robôs aquáticos realizarem algumas tarefas, utilizando redes neurais geradas a partir do algoritmo evolutivo NEAT (Neuroevolution of Augmenting Topologies) [8].…”

“…No trabalho de Duarte et al [3], criou-se um controle para agrupar os agentes ao redor de uma determinada coordenada GPS. A ideia deste trabalhoé semelhante, mas com a diferença de que essas coordenadas, aqui representadas pelo AV, mudam com o tempo.…”

Evolutionary Policy Transfer and Search Methods for Boosting Behavior Quality: RoboCup Keep-Away Case Study

“…Previous work has indicated that controllers evolved with NS function in a range of tasks of varying complexity (Velez and Clune, 2014) and such controllers consistently out-performed controllers evolved with objectivebased search in a range of ER tasks (Mouret and Doncieux, 2012;Gomes et al, 2015Gomes et al, , 2016. However, related research suggests that for complex tasks such as collective behavior tasks associated with swarm robotics (Duarte et al, 2016) (defined by high dimensional, rugged, discontinuous, and deceptive fitness landscapes (Eiben and Smith, 2003)), that evolutionary search hybridizing objectives (fitness functions) and NS tend to evolve effective high-quality solutions (behaviors) overall (Gomes and Christensen, 2013b;Gomes et al, 2013Gomes et al, , 2015. For a comprehensive survey of behavioral diversity maintenance methods used in various ER studies, the reader is referred to the review of Doncieux and Mouret (2014).…”

Abstraction as a Mechanism to Cross the Reality Gap in Evolutionary Robotics

“…With the use of a closed loop low-level control system, evolution would abstract away from the low-level sensor inputs and actuator outputs. Some resent work has shown promising results in bridging the reality gap using abstracted methods [5,7]. This however may come at a price, as abstraction "hides" the properties of the raw sensory inputs and motor actions to the controller, it may have as a disadvantage that the potential for SMC by the robot is reduced.…”