Control Barrier Function Constrained Model Reference Adaptive Control for UGV Under State and Input Limits

This paper studies constrained model reference adaptive control (MRAC) for a planar unmanned ground vehicle (UGV) subject to actuator limits and safety requirements. First, we establish a double-integrator model by applying dynamic feedback linearization to a nonholonomic kinematic model with acceleration input, while simultaneously accounting for external disturbances. A constrained MRAC scheme is developed that enforces constraints at two levels: (i) actuator constraints are guaranteed by saturating the physical inputs after mapping the adaptive virtual control through the inverse kinematic transformation, and (ii) safety constraints are enforced via componentwise control barrier function (CBF) on the tracking error, which induces explicit bounds on the plant state. A projection-based adaptive law is introduced to keep parameter estimates bounded and to ensure well-posedness under saturation-induced mismatch. Moreover, we propose a sufficient feasibility condition that explicitly relates safety margins, disturbance bounds, and available actuator authority, thereby forming a guideline for feasible region design. Simulation studies demonstrate that the proposed method achieves constraint-satisfying tracking under bounded disturbances while respecting physical actuator constraints.