Resumen:
We present our work on the training of robotised architectural components of intelligent buildings, focusing on main architectural components and features such as façades, roofs and partitions. The parameters governing such components may be either quantitative (such as temperature, humidity, configuration of the elements) or qualitative (such as ergonomics and aesthetics), which cannot easily be described by mathematical parameters. Due to their complexity,it is often impossible -or at least impractical, to hardcode suitable controllers for such robotised structures. Thus, we propose the use of Artificial Intelligence learning techniques, concretely Evolutionary Algorithms, so that the user can teach the robotised components how to behave in response to changing environmental conditions or user preferences. This idea is tested on an intelligent rooftop with variable geometry, that learns optimal configurations with respect to ambient light during training sessions.