Towards an Evolutionary Design of Modular Robots for Industry - Foundations on Natural and Artificial Computation - page 53

Fig. 1. Three of the modules of the architecture. Design (left) and prototype (right)
distributed control is possible. For the sake of simplicity, in this work we have
chosen to use a distributed control scheme where each module can control the
parameters of a sinusoidal that determines the motion of its actuator.
3 Evolutionary Design System
As indicated in the introduction, the problem we face here is that of developing
an automatic design system that provides an appropriate morphology for a task
or tasks together with an adequate control system. The key point in modular
systems is the desired degree of autonomy. At the higher level, the optimum
one, we have auto-reconfigurable adaptive systems, in which the control system
has some kind of learning algorithm that allows the modular system to achieve
new configurations on-line, or at least to adapt the current one to the changing
circumstances. In addition, the control system must be capable of detecting the
necessity of changing the task, and consequently, the configuration. This is a
highly complex problem that is currently in an early stage of development in
the field. As a first step to achieve such level of autonomy, here we are going
to address the problem of obtaining the control system and morphology for
each particular task in an off-line stage but through a general procedure, as
commented above.