“…Quite a few approaches have been proposed to sidestep the reality gap, through, e.g., closing the gap [3] or using dimensionality reduction [4]. On the other hand, the presented study resumes an earlier work, devoted to building low-cost robots [5]: there is no reliable simulator available for these robots, and it makes sense to design a training process relevant to such simulator-less experimental frameworks.…”
Section: Introductionmentioning
confidence: 80%
“…We believe that details of our physical robots are important as we tackle an on-board optimization of the controller, Resuming an earlier work [5], each robot ( Fig. 1) is equipped with 5 infrared (IR) proximity sensors, a camera, a lowcost micro-processing unit (MPU), and a LED light signaling its presence to the other robots.…”
Section: B Robot Architecturementioning
confidence: 99%
“…The robot locomotion uses two wheels and a supporting ball, enforcing a sufficient robustness of the robot (in contrast to our former use of two caterpillar [5]). The driving system consists of two motors each of which is connected to a wheel, a controlling circuit, a gear box, and the supporting ball.…”
Section: B Robot Architecturementioning
confidence: 99%
“…Note that we use a simplified version of the controllers used in [5], but we evolve it using the (1+1) ROAA [7], which we explain in the next section, during the experiments.…”
Section: A Reactive Controllermentioning
confidence: 99%
“…On the one hand, it takes much longer than using the simulator; 12 controller architectures have been considered in [5], and training each one of them in order to select the best one takes a large amount of time. On the other hand, in-situ training is physically demanding for the robots and might entail physical hazards.…”
Abstract-This paper aims at building autonomous controllers for swarm robots, specifically aimed at enforcing a given shape formation, here a column formation. The proposed approach features two main characteristics. Firstly, a state-of-the-art evolutionary setting is used to achieve the on-board optimization of the controller, avoiding any simulator-based approach. Secondly, as the cost of physical experiments might be prohibitively high for plain evolutionary approaches, a data mining approach is achieved on the top of evolution; rule discovery is used to discover the most promising regions in the controller search space. The merits of the approach are experimentally validated using a 5 robot formation, showing that the hybrid evolutionary learning process outperforms evolution alone in terms of swarm speed and shape quality.
“…Quite a few approaches have been proposed to sidestep the reality gap, through, e.g., closing the gap [3] or using dimensionality reduction [4]. On the other hand, the presented study resumes an earlier work, devoted to building low-cost robots [5]: there is no reliable simulator available for these robots, and it makes sense to design a training process relevant to such simulator-less experimental frameworks.…”
Section: Introductionmentioning
confidence: 80%
“…We believe that details of our physical robots are important as we tackle an on-board optimization of the controller, Resuming an earlier work [5], each robot ( Fig. 1) is equipped with 5 infrared (IR) proximity sensors, a camera, a lowcost micro-processing unit (MPU), and a LED light signaling its presence to the other robots.…”
Section: B Robot Architecturementioning
confidence: 99%
“…The robot locomotion uses two wheels and a supporting ball, enforcing a sufficient robustness of the robot (in contrast to our former use of two caterpillar [5]). The driving system consists of two motors each of which is connected to a wheel, a controlling circuit, a gear box, and the supporting ball.…”
Section: B Robot Architecturementioning
confidence: 99%
“…Note that we use a simplified version of the controllers used in [5], but we evolve it using the (1+1) ROAA [7], which we explain in the next section, during the experiments.…”
Section: A Reactive Controllermentioning
confidence: 99%
“…On the one hand, it takes much longer than using the simulator; 12 controller architectures have been considered in [5], and training each one of them in order to select the best one takes a large amount of time. On the other hand, in-situ training is physically demanding for the robots and might entail physical hazards.…”
Abstract-This paper aims at building autonomous controllers for swarm robots, specifically aimed at enforcing a given shape formation, here a column formation. The proposed approach features two main characteristics. Firstly, a state-of-the-art evolutionary setting is used to achieve the on-board optimization of the controller, avoiding any simulator-based approach. Secondly, as the cost of physical experiments might be prohibitively high for plain evolutionary approaches, a data mining approach is achieved on the top of evolution; rule discovery is used to discover the most promising regions in the controller search space. The merits of the approach are experimentally validated using a 5 robot formation, showing that the hybrid evolutionary learning process outperforms evolution alone in terms of swarm speed and shape quality.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.