Among the various possible criteria guiding eye movement selection, we investigate the role of position uncertainty in the peripheral visual field. In particular, we suggest that, in everyday life situations of object tracking, eye movement selection probably includes a principle of reduction of uncertainty. To evaluate this hypothesis, we confront the movement predictions of computational models with human results from a psychophysical task. This task is a freely moving eye version of the multiple object tracking task, where the eye movements may be used to compensate for low peripheral resolution. We design several Bayesian models of eye movement selection with increasing complexity, whose layered structures are inspired by the neurobiology of the brain areas implied in this process. Finally, we compare the relative performances of these models with regard to the prediction of the recorded human movements, and show the advantage of F. Colas (B) · F. Flacher · B. Girard
La mise au point de mécanismes de coordination spatiale pour des agents évoluant dans des univers continus et dynamiques est un problème difficile. Alors que la démarche descendante ne parvient pas à appliquer sa méthode de décomposition de façon satisfaisante sur cette classe de problèmes, l'approche ascendante obtient des résultats plus convaincants, mais elle implique souvent de fastidieux réglages manuels qui posent des problèmes de passage à l'échelle. Notre démarche pour traiter cette difficulté consiste à adjoindre à un formalisme de coordination spatiale ascendante un algorithme évolutionniste multicritère dédié à ce type de problèmes. Nous montrons sur un problème de coordination spatiale traité précédemment par Balch et Hybinette que les solutions obtenues avec notre plate-forme, GACS, sont comparables à celles obtenues par ces auteurs, malgré un investissement moindre de la part du concepteur. De plus, les solutions compétitives obtenues avec GACS sont plus simples que celles proposées par Balch et Hybinette, ce qui nous permet de conclure à la supériorité de notre approche.ABSTRACT. The design of spatial coordination mechanisms for dynamical and continuous multiagent setting is a difficult challenge. While the top-down decomposition approach is inefficient on such problems, the bottom-up approach is more promising, but requires a tedious manual parameter tuning which raises scaling-up issues. Our own approach consists in replacing the manual tuning by a specially designed multicriteria evolutionary algorithm devoted to the tuning of our spatial coordination formalism. In this paper, through a quantitative comparison on a complex spatial coordination problem treated previously by Balch and Hybinette, we show that our system, GACS, finds a population of solutions as efficient as this predecessor though our approach requires less involvement from the designers and can find simpler solutions.
Our approach to the spatial coordination problem relies on parametrized force fields. Through a quantitative comparison on a complex spatial coordination problem treated with a similar approach by Balch and Hybinette, we show that our system, GACS, finds a population of solutions as efficient as the ones found through an evolutionary optimisation of their handcrafted solution. Moreover GACS generates simpler solutions and requires much less involvement from the designer.
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.