Abstract
Robots have to move and survive in a variety of environments. In order to do so they have to avoid obstacles, find a collision free trajectory and cope with erroneous information; therefore, they must have rather efficient feedback loops, very similar to animal reflexes (Brooks, 1986). Real robots, however, have also to accomplish useful tasks, which are likely to have a reasonable degree of complexity. The purpose of this research is to perform experiments, in an unfriendly environment, with a robot having to accomplish tasks, which from a computational point of view may be rather complex, such as the exploration of a maze, the recovery of its structure and the planning of some nearly optimal trajectories through the maze. The robot used in the experimentation moves in a laboratory where a maze was constructed by sticking black stripes on the floor. The vision system of the robot is composed of a single uncalibrated T.V. camera pointing to the floor. In order to obtain almost real time performances, the perceptual world of the robot has been simplified, and is composed of black stripes and silhouettes over a flat floor with reflexes and shadows. A network of three computers is used to control the system behaviour. The processing of images is quite simple:the gray levels of the image are binarized, edges are extracted, and edge chains are approximated with a collection of straight segments. The robot has to recognize the direction of black stripes and the presence of junctions and silhouettes. The recognition of the silhouettes is based on the analysis of edge segments and takes into account the deformations due to perspective projection. In order to process the visual information correctly the system requires the setting of several parameters. By using an automatic learning procedure, given the lighting conditions, the material of the floor and of the black stripes, it was possible after a period of supervised learning to obtain a set of parameters, which could be used rather reliably.
Chapter PDF
Similar content being viewed by others
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1993 Springer-Verlag London Limited
About this paper
Cite this paper
Martinengo, A., Campani, M., Torre, V. (1993). Complex Tasks and Robots. In: Gielen, S., Kappen, B. (eds) ICANN ’93. ICANN 1993. Springer, London. https://doi.org/10.1007/978-1-4471-2063-6_75
Download citation
DOI: https://doi.org/10.1007/978-1-4471-2063-6_75
Published:
Publisher Name: Springer, London
Print ISBN: 978-3-540-19839-0
Online ISBN: 978-1-4471-2063-6
eBook Packages: Springer Book Archive