Control
Farnaz Sabahi
Abstract
Robotic manipulators operating in safety-critical environments must simultaneously guarantee accurate trajectory tracking, disturbance rejection, and state safety in the presence of model uncertainties. Conventional approaches typically combine μ-synthesis with Control Barrier Functions (CBFs) in ...
Read More
Robotic manipulators operating in safety-critical environments must simultaneously guarantee accurate trajectory tracking, disturbance rejection, and state safety in the presence of model uncertainties. Conventional approaches typically combine μ-synthesis with Control Barrier Functions (CBFs) in a cascaded architecture, where an online safety filter is applied after controller synthesis. However, this sequential design introduces unnecessary conservatism and precludes a unified theoretical treatment of robustness and safety. To address this limitation, this paper proposes a Dynamic Safety-Augmented μ-Synthesis (DSA-μ) framework that embeds safety dynamics directly into the generalized plant prior to controller synthesis. Parametric uncertainties are modeled using Linear Fractional Transformations (LFTs), while a dynamic safety state converts barrier conditions into augmented performance outputs for D–K iteration. The resulting integrated synthesis procedure yields a controller that simultaneously guarantees robust stability, disturbance attenuation, and forward invariance of the safe set. Numerical studies on a 3-DOF planar manipulator demonstrate that the proposed DSA-μ framework reduces trajectory-tracking RMSE by 35.5% compared with a conventional cascaded μ–CBF architecture while maintaining a positive safety margin throughout the simulations under 20% parametric model uncertainty.