odNEAT: An Algorithm for Distributed Online, Onboard Evolution of Robot Behaviours

Silva, Fernando; Urbano, Paulo; Oliveira, Sancho; Christensen, Anders Lyhne

doi:10.7551/978-0-262-31050-5-ch034

Cited by 32 publications

(28 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several authors have addressed on-line evolution of robotic agent controllers in different contexts: adaptation to dynamically changing environments ( [5]), parameter tuning ( [6]), evolution of self-assembly ( [2]), communication ( [11]), phototaxis and navigation ( [8], [12]). Some of this work is detailed in the next section.…”

Section: Introductionmentioning

confidence: 99%

Comparison of Selection Methods in On-Line Distributed Evolutionary Robotics

Pérez¹,

Boumaza²,

Charpillet³

2014

Artificial Life 14: Proceedings of the Fourteenth International Conference on the Synthesis and Simulation of Living Systems

View full text Add to dashboard Cite

In this paper, we study the impact of selection methods in the context of on-line on-board distributed evolutionary algorithms. We propose a variant of the mEDEA algorithm in which we add a selection operator, and we apply it in a task-driven scenario. We evaluate four selection methods that induce different intensity of selection pressure in a multi-robot navigation with obstacle avoidance task and a collective foraging task. Experiments show that a small intensity of selection pressure is sufficient to rapidly obtain good performances on the tasks at hand. We introduce different measures to compare the selection methods, and show that the higher the selection pressure, the better the performances obtained, especially for the more challenging food foraging task.

show abstract

Section: Introductionmentioning

confidence: 99%

Comparison of Selection Methods in On-Line Distributed Evolutionary Robotics

Pérez¹,

Boumaza²,

Charpillet³

2014

Artificial Life 14: Proceedings of the Fourteenth International Conference on the Synthesis and Simulation of Living Systems

View full text Add to dashboard Cite

show abstract

“…Complete descriptions of the method are available in [9][10][11]. odNEAT was originally designed to run across a distributed group of agents whose objective is to evolve and adapt while operating in the environment.…”

Section: Odneat: An Online Evolutionary Algorithmmentioning

confidence: 99%

“…odNEAT [9] is an online, distributed and decentralised version of NEAT [10]. The NEAT method, one of the most prominent neuroevolution (NE) algorithms, is capable of optimising both the topology of the network and its connection weights.…”

Section: Odneat: An Online Evolutionary Algorithmmentioning

confidence: 99%

See 1 more Smart Citation

Adaptation of Robot Behaviour through Online Evolution and Neuromodulated Learning

Silva

Urbano

Christensen

2012

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Abstract. We propose and evaluate a novel approach to the online synthesis of neural controllers for autonomous robots. We combine online evolution of weights and network topology with neuromodulated learning. We demonstrate our method through a series of simulation-based experiments in which an e-puck-like robot must perform a dynamic concurrent foraging task. In this task, scattered food items periodically change their nutritive value or become poisonous. Our results show that when neuromodulated learning is employed, neural controllers are synthesised faster than by evolution alone. We demonstrate that the online evolutionary process is capable of generating controllers well adapted to the periodic task changes. An analysis of the evolved networks shows that they are characterised by specialised modulatory neurons that exclusively regulate the output neurons.

show abstract

“…This type of evolution is often referred to as environmentdriven evolution [2]. Typical approaches such as [14] remove the need for any central control, resulting in algorithms that perform distributed and online evolution. An additional feature of environment-driven algorithms is that no explicit fitness function is defined: instead, mate selection and reproduction depend on selection pressure provided only by the environment yet need to lead to stable populations.…”

Section: Introductionmentioning

confidence: 99%

Improving Survivability in Environment-driven Distributed Evolutionary Algorithms through Explicit Relative Fitness and Fitness Proportionate Communication

Hart

Steyven

Paechter

2015

Proceedings of the 2015 Annual Conference on Genetic and Evolutionary Computation

View full text Add to dashboard Cite

Ensuring the integrity of a robot swarm in terms of maintaining a stable population of functioning robots over long periods of time is a mandatory prerequisite for building more complex systems that achieve user-defined tasks. mEDEA is an environment-driven evolutionary algorithm that provides promising results using an implicit fitness function combined with a random genome selection operator. Motivated by the need to sustain a large population with sufficient spare energy to carry out user-defined tasks in the future, we develop an explicit fitness metric providing a measure of fitness that is relative to surrounding robots and examine two methods by which it can influence spread of genomes. Experimental results in simulation find that use of the fitness-function provides significant improvements over the original algorithm; in particular, a method that influences the frequency and range of broadcasting when combined with random selection has the potential to conserve energy whilst maintaining performance, a critical factor for physical robots.

show abstract

odNEAT: An Algorithm for Distributed Online, Onboard Evolution of Robot Behaviours

Cited by 32 publications

References 6 publications

Comparison of Selection Methods in On-Line Distributed Evolutionary Robotics

Comparison of Selection Methods in On-Line Distributed Evolutionary Robotics

Adaptation of Robot Behaviour through Online Evolution and Neuromodulated Learning

Improving Survivability in Environment-driven Distributed Evolutionary Algorithms through Explicit Relative Fitness and Fitness Proportionate Communication

Contact Info

Product

Resources

About