Layout optimization is essential to build a multi-robot system. Layout parameters affect the cooperation efficiency and the utilization rate of the workspace. This paper conducts layout optimization of a dual-robot system for mirror milling. It contains a hybrid robot for milling and a serial robot for supporting during the machining of thin-walled parts. The milling robot and supporting robot are modeled geometrically, and the reachable workspaces of the two robots are obtained with the help of forward and inverse kinematics. An algorithm for solving the maximum regular workspace based on a nonlinear programming method is developed, which does not require complex mathematical derivation and is universal for both concave and convex polygons. The manipulability analysis of the serial robot at the support end is carried out, and the area with low manipulability is eliminated from the reachable space. A layout optimization method that involves the reachability, workspace regularity, and manipulability of the robot is proposed and validated by theoretical analyses.