Adaptive robot control gained by partial identification using the advantages of symplectic geometry

摘要

The aim of this paper is to reveal the advantages of using the tools of the "modern formalism" of "classical analytical mechanics" in the control of imperfectly modeled, coupled nonlinear mechanical systems having dynamic interaction with an unmodelled environment. It is shown that the phenomenological difficulties inherent in this description and leading to its ignorance in robot control can be bypassed by a higher level of abstraction resulting in hardware and/or software blocks of uniform structure. The great number of arbitrary parameters of the new method guarantees versatility and also leads to open problems investigated by numerical simulations in the paper.