Skip to Main Content
Nonlinear proportional-integral-derivative (NPID) is realized by modulating the control gains of a conventional linear controller in response to system state or error. Performance benefits such as increased damping or reduced tracking error have been experimentally demonstrated. Previous results (2000) have established Lyapunov stability of NPID control for full-state knowledge and for special cases of partial-state knowledge; but the general problem of stability with arbitrary state knowledge remained unsolved. In the present work, design methods for NPID control are extended to the case of arbitrary partial-state knowledge. Given any combination of a stabilizable linear system and NPID control law, an upper bound on control effort can be determined for which global asymptotic stability is assured. The result is based on a Lyapunov stability proof and is applied to establish the stability of NPID control for the Sarcos dexterous manipulator.