This paper introduces a novel real-time geometric calculation method designed to significantly enhance the control accuracy of the end effector (EE) in Mobile Cable-Driven Parallel Robots (MCDPRs) operating on uneven arrangement. Utilizing eight cables, the MCDPR facilitates the control of the EE's 6-DOF and force, while four mobile platforms serve as bases, enabling free movement and deformation of the geometry. The essence of this method lies in the fusion of wheel encoder and IMU data to achieve precise 3D localization, which forms the basis for accurate geometric calculations even on uneven grounds. Localization relies on three-dimensional odometry, with quaternion calculations from IMU data and complementary filtering used to determine the precise pose of each mobile platform. Furthermore, to minimize yaw direction errors caused by wheel slip, a fuzzy filter was employed to calculate the yaw rate accurately. Geometry and cable length calculations for EE control in MCDPR were conducted based on localization data, using Homogeneous Transformation Matrices (HTM) for the robot's key components. The method’s effectiveness was validated through experiments on EE control within an uneven MCDPR arrangement. The EE was controlled to follow a path along a multi-layer circular structure, and the results demonstrated that errors remained within 5% of the path radius, irrespective of the MCDPR configuration, thereby confirming the superior performance of the geometric calculation method.