A carregar...
Projeto de investigação
Sem título
Financiador
Autores
Publicações
Novel approaches to cooperative coevolution of heterogeneous multiagent systems
Publication . Gomes, Jorge Miguel Carvalho; Christensen, Anders Lyhne; Mariano, Pedro, 1975-; Correia, Luís Miguel Parreira e, 1959-
Heterogeneous multirobot systems are characterised by the morphological and/or behavioural heterogeneity of their constituent robots. These systems have a number of advantages over the more common homogeneous multirobot systems: they can leverage specialisation for increased efficiency, and they can solve tasks that are beyond the reach of any single type of robot, by combining the capabilities of different robots. Manually designing control for heterogeneous systems is a challenging endeavour, since the desired system behaviour has to be decomposed into behavioural rules for the individual robots, in such a way that the team as a whole cooperates and takes advantage of specialisation. Evolutionary robotics is a promising alternative that can be used to automate the synthesis of controllers for multirobot systems, but so far, research in the field has been mostly focused on homogeneous systems, such as swarm robotics systems. Cooperative coevolutionary algorithms (CCEAs) are a type of evolutionary algorithm that facilitate the evolution of control for heterogeneous systems, by working over a decomposition of the problem. In a typical CCEA application, each agent evolves in a separate population, with the evaluation of each agent depending on the cooperation with agents from the other coevolving populations. A CCEA is thus capable of projecting the large search space into multiple smaller, and more manageable, search spaces. Unfortunately, the use of cooperative coevolutionary algorithms is associated with a number of challenges. Previous works have shown that CCEAs are not necessarily attracted to the global optimum, but often converge to mediocre stable states; they can be inefficient when applied to large teams; and they have not yet been demonstrated in real robotic systems, nor in morphologically heterogeneous multirobot systems. In this thesis, we propose novel methods for overcoming the fundamental challenges in cooperative coevolutionary algorithms mentioned above, and study them in multirobot domains: we propose novelty-driven cooperative coevolution, in which premature convergence is avoided by encouraging behavioural novelty; and we propose Hyb-CCEA, an extension of CCEAs that places the team heterogeneity under evolutionary control, significantly improving its scalability with respect to the team size. These two approaches have in common that they take into account the exploration of the behaviour space by the evolutionary process. Besides relying on the fitness function for the evaluation of the candidate solutions, the evolutionary process analyses the behaviour of the evolving agents to improve the effectiveness of the evolutionary search. The ultimate goal of our research is to achieve general methods that can effectively synthesise controllers for heterogeneous multirobot systems, and therefore help to realise the full potential of this type of systems. To this end, we demonstrate the proposed approaches in a variety of multirobot domains used in previous works, and we study the application of CCEAs to new robotics domains, including a morphological heterogeneous system and a real robotic system.
Evolutionary online behaviour learning and adaptation in robotic systems
Publication . Silva, Fernando Goulart da; Correia, Luís Miguel Parreira e, 1959-; Christensen, Anders Lyhne
In this thesis, we study new ways to enable ecient online learning in autonomous robots. We employ a control synthesis methodology called evolutionary robotics, which emerged in the 1990s as a promising alternative to classic articial intelligence techniques and design methodologies for control systems. In online learning through evolution, henceforth called online evolution, an evolutionary algorithm is executed onboard each robot in order to create and continuously optimise its behavioural control logic. Each instance of the evolutionary algorithm executes without any external supervision or human intervention. Online evolution can thus automatically generate the articial intelligence that controls each robot, and creates the potential for long-term behaviour adaptation and learning: robots can continuously self adjust and learn new behaviours in response to, for example, changes in the task requirements or environmental conditions, and to faults in the sensors and/or actuators. Despite the potential for automatic behaviour learning, online evolution is not frequently employed for a number of reasons. First, online evolution typically requires several hours or days to synthesise solutions to a task. As a result, the approach has not yet been practically exploited in real-robot systems. Second, one common assumption in the eld is that online evolution enables continuous learning and adaptation to previously unforeseen circumstances. However, only a small number of ad-hoc experiments have been carried out in simulation. That is, the potential for online evolution to enable dynamic adaptation and learning has been largely left unstudied. The main goal of this thesis is to address some of the fundamental issues associated with online evolution to bring it closer to widespread adoption. Our research focuses on studying if and how to accelerate and increase the performance of online evolution. Our rst research contribution is a comprehensive presentation and analysis of Online Decentralised NeuroEvolution of Augmenting Topologies (odNEAT), an algorithm for online evolution of neural networkbased controllers in multirobot systems. odNEAT diers from more traditional approaches to online evolution because both the weighting parameters and the topological structure of neural networks are under evolutionary control. Our second research contribution focuses on investigating the dynamics of online evolution of controllers at two dierent levels. At the microscopic scale, we assess the dynamics of distinct neuronal models. At the macroscopic scale, we investigate the scalability properties of online evolution with respect to group size. The outcomes of the contribution are an assertion of the critical role of the controller evaluation policy, and an analysis of how the group size inuences task performance. In our third research contribution, we capitalise on the knowledge gained from the second study, and we introduce: (i) a racing approach that allows individual robots to cut short the evaluation of poor controllers, (ii) a population cloning approach that enables each individual robot to clone and transmit a varying number of high-performing controllers to other robots nearby, and (iii) online hyper-evolution (OHE), an unprecedented approach in evolutionary robotics with the capability to automatically construct algorithms for controller generation during task execution. To conclude, we validate our research in real robotic hardware, and we successfully demonstrate evolution of controllers to solve three classic evolutionary robotics tasks in a timely manner (one hour or less).
Unidades organizacionais
Descrição
Palavras-chave
Contribuidores
Financiadores
Entidade financiadora
Fundação para a Ciência e a Tecnologia
Programa de financiamento
5876
Número da atribuição
UID/EEA/50008/2013