German Title: Optimierung der Stabilitaet von Laufrobotern ohne Feedback

	Postscript, English Download (6MB) | Terms of use
Preview	PDF, English Download (4MB) | Terms of use

Translation of abstract (English)

This thesis presents fundamentally new open-loop stable one-legged and two-legged walking robots. It thus demonstrates that the concept of open-loop control is applicable to a much wider range of systems than was conceived before. An open-loop stable control strategy does not use any feedback nor does it take any active measures to respond to perturbations but entirely relies on the mechanical system's natural kinematics and dynamics to stabilize the trajectory. Its outstanding advantages are low cost an speed of control. The design of such open-loop stable walking robots and the choice of model parameters leading to stable motions is a difficult task and has not been much investigated before. In this thesis it has been accomplished by means of optimization. A numerical method for the optimization of open-loop stability of periodic systems has been developed. This is the first time stability optimization is combined with the simultaneous solution of an optimal control problem. The problem of periodic gait generation is addressed as a subtask. Robots are modeled as multi-body systems involving discontinuities. Besides open-loop controlled actuated robots also the special case of passive-dynamic walkers is considered, that, in addition to lacking feedback control, have no active sources of energy. The algorithm delivered excellent results for all robot models studied and could be applied to more general periodic dynamical systems.