Modern Compliant Robot Control: Exploring Benefits From Singular Perturbation Synthesis

Singular perturbation (SP)-based control is promising for flexible joint robots (FJRs) to achieve high tracking accuracy under low implementation cost, and composite learning can achieve exact parameter estimation without persistent excitation. To exploit the benefits of SP-based control synthesis in the tracking control problem for FJRs with high degrees of freedom (DoFs), this article introduces five FJR controllers, including SP-based and non- SP-based controllers, for modern compliant robots with high DoFs and nonnegligible joint elasticity. A unique and in-depth comparative analysis of these controllers is provided to reveal their technical natures. It is shown that SP-based controllers have significant advantages for tracking control of FJRs, and composite learning is useful for SP-based controllers to achieve superior tracking under model uncertainties. Simulation and experimental comparisons based on a compliant robot with seven DoFs named Franka Emika Panda have verified the above arguments. Moreover, quantitative comparisons have revealed the relationship between the control and estimation performance of the five controllers and inner-loop control gains.