Abstract: During the operation of unmanned mining dump trucks in open-pit mining environments, significant actuator delays are commonly observed, which lead to reduced trajectory tracking accuracy and degraded operational stability. To address this issue, a trajectory tracking control strategy that explicitly accounts for actuator delay is proposed. First, a composite model consisting of a first-order inertial element and a pure time-delay element is employed to accurately characterize the delay dynamics of both longitudinal and lateral actuators. Based on this modeling framework, a longitudinal dynamic model and a lateral trajectory-tracking kinematic error model are established. On this basis, a decoupled longitudinal–lateral control architecture is adopted, within which a longitudinal Model Predictive Control (MPC) scheme and a lateral Nonlinear Model Predictive Control (NMPC) scheme are formulated. The objective functions incorporate trajectory tracking accuracy, operational efficiency, and driving safety, while actuator performance constraints are explicitly imposed to ensure the feasibility of control inputs and overall system stability. Furthermore, a bidirectional vehicle-state interaction mechanism over the prediction horizon is introduced into the decoupled framework to achieve coordinated control between the longitudinal and lateral controllers. The proposed method is validated through both hardware-in-the-loop (HIL) simulations and real-vehicle experiments based on actual road scenarios in an open-pit coal mine, with a Komatsu 930E mining dump truck selected as the test platform. Comparative evaluations are conducted against a field-deployed onboard controller currently used in the mine. The real-vehicle experimental results indicate that, on the same road segment, the proposed controller reduces the peak lateral error, peak heading error, and peak velocity tracking error by 61.3%, 57.1%, and 35.1%, respectively. In addition, it demonstrates clear advantages in trajectory tracking accuracy and operational smoothness, while improving operational efficiency by approximately 9.6%. These results verify the effectiveness and superiority of the proposed controller in practical applications.