<p>Mobile robotic milling systems offer high flexibility and cost-effectiveness for machining distributed features of large-scale components. However, the milling stability and stiffness performance of mobile industrial robots are inherently limited and vary nonlinearly with robot pose across a large workspace, which presents significant challenges to achieving high-precision machining. To address these issues, this paper proposes a data–mechanism hybrid-driven pose optimization method that accounts for both milling stability and stiffness performance. A data-driven milling stability model is developed using AdaBoost-Gaussian Process Regression (AdaBoost-GPR) combined with the zero-order analytical (ZOA) method, effectively mitigating the low accuracy in modal parameter prediction caused by strong nonlinear pose-dependent variations. Meanwhile, the joint stiffness of the ESTUN ER350-3300 robot is identified, and a stiffness model is established by mapping compliance in the cutting direction. A comprehensive index is then introduced by integrating the milling stability and stiffness models, and an Energy Valley Optimizer (EVO)-based algorithm is employed to optimize this index. The optimization process incorporates kinematic constraints and mode-coupling chatter as key considerations. Finally, milling experiments under various poses and machining parameters validate the proposed approach, demonstrating improved milling stability, enhanced stiffness performance, and a more reliable machining process for mobile robotic milling systems.</p>