The purpose of this project is the exploration of how autonomous robots could develop a language to communicate visual patterns. The idea is that each robot should be able to change their visual appearance depending on what its neighbors are trying to communicate. Thus, the robots should `talk' about their patterns, trying to influence each other. For this project we used the e-Puck robot, a small mobile robot developed by EPFL (Ecole Polytechnique F\'{e}d\'{e}rale de Lausanne) in conjunction with the see-Puck display. The display, developed by FAL (Future Applications Lab), consists of a matrix of 148 LEDs in a circular shape. During this project we looked into several methods of achieving communication though the sensors and actuators of the e-Puck robot. An additional area which was explored was the process of user interaction with the robots.

Missing pages were blank, therefore omitted. / Practice improves sensory perception, a phenomenon known as perceptual learning. Perceptual learning is interesting because it reflects plasticity in the brain where none was imagined, and because of its enormous applied potential. In vision, learning of simple discriminations is well-described. Here, I study the learning of two complex visual tasks, texture-and face identification, using a ten-alternative forced-choice procedure. The data are clear: learning of complex patterns is much like learning of simple patterns in its specificity, stability and time-course. Therefore, learning obeys similar rules at several levels in visual processing. The characteristics of learning, in particular the specificity and stability of learning, affect inherent aspects of object recognition. / Thesis / Doctor of Philosophy (PhD)