Racing to improve on-line, on-board evolutionary robotics

Haasdijk, Evert; Arif, Atta-ul-Qayyum; Eiben, A. E.

doi:10.1145/2001576.2001603

Cited by 12 publications

(5 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Swarms of robots are ideal for tasks where large areas need to be covered for monitoring or surveillance purposes. This task category has been explored in the past with several different control techniques, such as pheromone traces [70], odor sources [71] and evolved artificial neural networks [72]. In our area monitoring task, a geo-fence is defined to delimit an area of interest, and the robots should coordinate to continuously cover as much of the area as possible.…”

Section: Area Monitoringmentioning

confidence: 99%

See 1 more Smart Citation

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

et al. 2016

View full text Add to dashboard Cite

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.

show abstract

Section: Area Monitoringmentioning

confidence: 99%

Section: Area Monitoringmentioning

confidence: 99%

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

et al. 2016

View full text Add to dashboard Cite

show abstract

“…In irace, subsequent races re-use information from previous races and the best (elites) solutions found can only be replaced by strictly better solutions. Although racing is far less studied than other methods for handling uncertainty in the evolutionary optimization literature [17], its use is becoming increasingly widespread [2,12,13,14,33]. Our results show that irace is able to obtain optimized traffic-light programs with much lower fitness and variance than those obtained by the GA with any of the resampling strategies.…”

Section: Introductionmentioning

confidence: 83%

Reliable simulation-optimization of traffic lights in a real-world city

Ferrer

López-Ibáñez

Alba

2019

Applied Soft Computing

View full text Add to dashboard Cite

In smart cities, when the real-time control of traffic lights is not possible, the global optimization of traffic-light programs (TLPs) requires the simulation of a traffic scenario (traffic flows across the whole city) that is estimated after collecting data from sensors at the street level. However, the highly dynamic traffic of a city means that no single traffic scenario is a precise representation of the real system, and the fitness of any candidate solution (traffic-light program) will vary when deployed on the city. Thus, ideal TLPs should not only have an optimized fitness, but also a high reliability, i.e., low fitness variance, against the uncertainties of the real-world. Earlier traffic-light optimization methods, e.g., based on genetic algorithms, often simulate a single traffic scenario, which neglects variance in the real-world, leading to TLPs not optimized for reliability.Our main contributions in this work are the following: (a) the analysis of the importance of reliable solutions for TLP optimization, even when all traffic scenarios are consistent with the real-world data and highly correlated; (b) the adaptation of irace, an iterated racing algorithm that is able to dynamically adjust the number of traffic scenarios required to evaluate the fitness of TLPs and their reliability; (c) the use of a large real-world case study for which real-time control is not possible and where data was obtained from sensors at the street level; and (d) a thorough analysis of solutions generated by means of irace, a Genetic Algorithm, a Differential Evolution, a Particle Swarm Optimization and a Random Search. This analysis shows that simple strategies that simulate multiple traffic scenarios are able to obtain optimized solutions with improved reliability; however, the best results are obtained by irace, among the algorithms evaluated.

show abstract

“…Several established methods exist to reduce the number of evaluations required or cut short an individual evaluation. Examples include as early stopping algorithms (Bongard, 2011) and racing techniques (Haasdijk et al, 2011). However, one of the goals of ER is to realize robots capable of operating in noisy or dynamic environments (Bongard, 2009), and that can execute multiple tasks in parallel or in sequence (Nolfi, 2002).…”

Section: Speeding Up Evolution In Real Hardwarementioning

confidence: 99%

Open Issues in Evolutionary Robotics

Silva

Duarte

Correia

et al. 2016

Evolutionary Computation

View full text Add to dashboard Cite

One of the long-term goals in evolutionary robotics is to be able to automatically synthesize controllers for real autonomous robots based only on a task specification. While a number of studies have shown the applicability of evolutionary robotics techniques for the synthesis of behavioral control, researchers have consistently been faced with a number of issues preventing the widespread adoption of evolutionary robotics for engineering purposes. In this article, we review and discuss the open issues in evolutionary robotics. First, we analyze the benefits and challenges of simulation-based evolution and subsequent deployment of controllers versus evolution on real robotic hardware. Second, we discuss specific evolutionary computation issues that have plagued evolutionary robotics: (1) the bootstrap problem, (2) deception, and (3) the role of genomic encoding and genotype-phenotype mapping in the evolution of controllers for complex tasks. Finally, we address the absence of standard research practices in the field. We also discuss promising avenues of research. Our underlying motivation is the reduction of the current gap between evolutionary robotics and mainstream robotics, and the establishment of evolutionary robotics as a canonical approach for the engineering of autonomous robots.

show abstract

Racing to improve on-line, on-board evolutionary robotics

Cited by 12 publications

References 12 publications

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

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

Reliable simulation-optimization of traffic lights in a real-world city

Open Issues in Evolutionary Robotics

Contact Info

Product

Resources

About