[Objective] To address the issues of poor disturbance rejection performance and slow response speed exhibited by traditional proportional-integral-derivative (PID) control strategy for vehicle-mounted stabilized gimbals operating on unpaved roads, this paper proposes a single-neural adaptive (SNA)-PID control strategy based on a linear extended state observer (LESO). This approach aims to enhance the stability and disturbance rejection capability of the gimbal’s posture control. [Methods] Firstly, a mathematical model of the stepper motor was established. Then, based on the LESO principle and the supervised Hebb learning rule, a LESO-based SNA-PID control strategy was designed. The convergence of the proposed control strategy was verified using Lyapunov stability theory. Finally, to validate the effectiveness of the proposed strategy, a practical vehicle-mounted stabilization gimbal platform was constructed, and comparative experiments were conducted under both internal and external disturbances against conventional PID and LESO+PID control strategy. [Results] Compared with traditional PID and LESO+PID control, the proposed control strategy significantly reduced angle-tracking peak-to-peak values and other performance indices under single disturbance, 1 Hz continuous disturbance, and 8 Hz continuous disturbance conditions. Although a slight increase in overshoot was observed when the reference signal changed, the settling time was substantially reduced. [Conclusion] The proposed LESO+SNA-PID control strategy forms a dual-compensation closed-loop strategy, which effectively enhances the disturbance rejection capability against external disturbances and improves the system’s response speed, demonstrating significant value for practical engineering applications.