A simple neural network controller merging different behaviors for collector robots

Antoine de Rengervé (1) , Julien Hirel (1) , Mathias Quoy (1) , Pierre Andry (1) , Philippe Gaussier (1)

1 ETIS - UMR 8051 - Equipes Traitement de l'Information et Systèmes

Robots are expected to become reliable partners for working in human environments. One of the tasks that they could have to perform is collecting objects and gathering/sorting them for any purposes. In such a task the robot would have to explore its environment to discover objects of interest, navigate toward them, grasp them and navigate again to a storage area to drop the object. In this paper we show how the basic principles behind these complex behaviors can be obtained from fairly simple neural network controllers.

Domaines: Réseau de neurones [cs.NE]; Robotique [cs.RO]

Antoine de Rengervé, Julien Hirel, Mathias Quoy, Pierre Andry, Philippe Gaussier. A simple neural network controller merging different behaviors for collector robots. International Workshop on bio-inspired robots, Apr 2011, Nantes, France. ⟨hal-00631097⟩