Direct current (DC) servo motors are essential in industrial automation and robotics, where precise position and speed control are critical; however, they require high-performance control strategies. This study investigates an experimental framework for the modeling and control of a DC servo motor using system identification and quantitatively compares a traditional Ziegler–Nichols (ZN)-tuned proportional-integral-derivative (PID), a traditional linear quadratic regulator (LQR), and the proposed adaptive policy iteration on an identified model. A second-order transfer function derived from the closed-loop step-response data showed good agreement with the measured dynamics. Experimental results indicated that the real-time PID controller achieved the fastest rise time but exhibited a larger overshoot and weaker robustness to parameter variations compared with other strategies. The optimal LQR design provided improved damping and stability margins, significantly reducing the overshoot at the cost of a slightly slower response. The proposed adaptive-policy-iteration approach demonstrated superior adaptability, convergence of online near-optimal gains, reduced control effort, and improved trajectory tracking under varying conditions. The findings present the system-identification framework as a practical solution for deriving accurate models for industrial servo motors. Additionally, this study highlights trade-offs in tracking performance; the proposed adaptive policy iteration improves the transient response compared to traditional PID and LQR methods, but at the cost of a slightly increased overshoot.