Intelligent control for complex robotic systems

In order to give complex mechanical systems more capabilities in their dynamical behavior, the classical approach must be amended to comply with their structure. As they exhibit a complicated and mixed pattern of displacements and deformations, their trajectories become very complicated, and no structured information can be obtained from them. More global quantities, such as power input which fixes a set of possible trajectories in a fixed function space, must be determined. This leads to the concept of functional control which aims at controlling the system globally from its accessible previous rigid variables without entering the (unusable) detail of exact trajectories. Possible improvement is obtained by adding learning type control, and by setting parameters so that a converging fixed point property can be obtained. An upper control loop of a more qualitative decisional nature can be added on top of a functional low level loop, freeing the choice of system behavior in an unstructured environment