2009 Ieee Rsj International Conference on Intelligent Robots and Systems, Oct 1, 2009
We study a simple algorithm inspired by the Brazil nut effect for achieving segregation in a swar... more We study a simple algorithm inspired by the Brazil nut effect for achieving segregation in a swarm of mobile robots. The algorithm lets each robot mimic a particle of a certain size and broadcast this information locally. The motion of each particle is controlled by three reactive behaviors: random walk, taxis, and repulsion by other particles. The segregation task requires
Proceedings of the 2014 conference on Genetic and evolutionary computation - GECCO '14, 2014
ABSTRACT We propose a coevolutionary approach for learning the behavior of animals, or agents, in... more ABSTRACT We propose a coevolutionary approach for learning the behavior of animals, or agents, in collective groups. The approach requires a replica that resembles the animal under investigation in terms of appearance and behavioral capabilities. It is able to identify the rules that govern the animals in an autonomous manner. A population of candidate models, to be executed on the replica, compete against a population of classifiers. The replica is mixed into the group of animals and all individuals are observed. The fitness of the classifiers depends solely on their ability to discriminate between the replica and the animals based on their motion over time. Conversely, the fitness of the models depends solely on their ability to 'trick' the classifiers into categorizing them as an animal. Our approach is metric-free in that it autonomously learns how to judge the resemblance of the models to the animals. It is shown in computer simulation that the system successfully learns the collective behaviors of aggregation and of object clustering. A quantitative analysis reveals that the evolved rules approximate those of the animals with a good precision.
ABSTRACT This paper presents a multi-robot solution to the task of object clustering, where the s... more ABSTRACT This paper presents a multi-robot solution to the task of object clustering, where the simplicity of the robots is pushed to the extreme that (i) each robot can only detect the presence of (but not the distance to) an object or another robot in its direct line of sight, and (ii) the robots are unable to store previous inputs and cannot perform arithmetic computations. Controllers for the robots were synthesized through an evolutionary robotics approach driven by physics-based simulations. The results show that the problem can be solved even if the robots cannot distinguish between objects and other robots; however, if they are able to make this distinction, the clustering performance is significantly improved. The controllers have been shown to scale well to large numbers of robots and objects and to be robust to noise. The sensor/controller solution was implemented on the e-puck robotic system. Across 10 systematic experiments with 5 robots and 20 objects, on average, 86.5% of the objects were in one cluster after 10 minutes. We believe that the sensor/controller simplicity paves the way for the implementation of multi-robot systems at very small scales, as required, for instance, in nanomedical applications.
2009 Ieee Rsj International Conference on Intelligent Robots and Systems, Oct 1, 2009
We study a simple algorithm inspired by the Brazil nut effect for achieving segregation in a swar... more We study a simple algorithm inspired by the Brazil nut effect for achieving segregation in a swarm of mobile robots. The algorithm lets each robot mimic a particle of a certain size and broadcast this information locally. The motion of each particle is controlled by three reactive behaviors: random walk, taxis, and repulsion by other particles. The segregation task requires
Proceedings of the 2014 conference on Genetic and evolutionary computation - GECCO '14, 2014
ABSTRACT We propose a coevolutionary approach for learning the behavior of animals, or agents, in... more ABSTRACT We propose a coevolutionary approach for learning the behavior of animals, or agents, in collective groups. The approach requires a replica that resembles the animal under investigation in terms of appearance and behavioral capabilities. It is able to identify the rules that govern the animals in an autonomous manner. A population of candidate models, to be executed on the replica, compete against a population of classifiers. The replica is mixed into the group of animals and all individuals are observed. The fitness of the classifiers depends solely on their ability to discriminate between the replica and the animals based on their motion over time. Conversely, the fitness of the models depends solely on their ability to 'trick' the classifiers into categorizing them as an animal. Our approach is metric-free in that it autonomously learns how to judge the resemblance of the models to the animals. It is shown in computer simulation that the system successfully learns the collective behaviors of aggregation and of object clustering. A quantitative analysis reveals that the evolved rules approximate those of the animals with a good precision.
ABSTRACT This paper presents a multi-robot solution to the task of object clustering, where the s... more ABSTRACT This paper presents a multi-robot solution to the task of object clustering, where the simplicity of the robots is pushed to the extreme that (i) each robot can only detect the presence of (but not the distance to) an object or another robot in its direct line of sight, and (ii) the robots are unable to store previous inputs and cannot perform arithmetic computations. Controllers for the robots were synthesized through an evolutionary robotics approach driven by physics-based simulations. The results show that the problem can be solved even if the robots cannot distinguish between objects and other robots; however, if they are able to make this distinction, the clustering performance is significantly improved. The controllers have been shown to scale well to large numbers of robots and objects and to be robust to noise. The sensor/controller solution was implemented on the e-puck robotic system. Across 10 systematic experiments with 5 robots and 20 objects, on average, 86.5% of the objects were in one cluster after 10 minutes. We believe that the sensor/controller simplicity paves the way for the implementation of multi-robot systems at very small scales, as required, for instance, in nanomedical applications.
Uploads