To address the limitations of single sensor perception and the strong coupling of motion control in unmanned vehicles in complex dynamic scenarios, this paper conducts research on an environmental perception and control system based on SLAM. Design a multi-sensor fusion SLAM architecture, achieve robust state estimation and high-precision map construction through the tight coupling of lidar, camera and IMU, and introduce point cloud denoising and time-sequence synchronization mechanisms to improve perception quality. A hierarchical path planning strategy is proposed, and a transverse and longitudinal cooperative motion controller based on model predictive control is designed, fully considering the vehicle dynamic constraints and road boundaries. Real vehicle tests show that the system achieves centimeter-level positioning accuracy in complex scenarios, with a lateral tracking error of no more than 18.3cm, which can provide reliable technical support for autonomous navigation of unmanned vehicles.