This paper provides a path planning system design for autonomous haulage applications through a Model-Based System Engineering (MBSE) approach. In the autonomous haulage system (AHS) of mining, the processes of loading, hauling, and unloading can be decomposed into multiple trajectory-following tasks with predefined trajectories and target positions. Supported by the perception, path planning, and vehicle control subsystems, the autonomous truck automatically follows the predefined trajectory to complete the production tasks scheduled by the central server. The path planning system provides collision-free, smooth, and efficient trajectories for the trucks to track. In cases with unknown obstacles and changing environments, the system offers re-planning features to ensure safety and efficiency. A four-level design scheme for a path planning system is obtained through a systems engineering process using the SysML and EAST-ADL architectures. This design allows engineers to build systems suitable for semi-structured environments, balancing both adaptability and performance.