<p>Heterogeneous Multi-Robot Cable-Driven Parallel Manipulator Systems (MCDPM) represent a specific type of system that uses aerial and mobile vehicles to manipulate an end effector platform. Planning paths for such systems involves complex challenges such as coordination, scalability, consideration of uncertainty, communication limitations and resource constraints. There are no known non-planar search-based path planning strategies for these systems that account for constraints, such as cable interference and stability. This article proposes a novel approach for such systems that considers 3D spatial variations and generates real-time, flexible formation plans. The path of the end effector platform is determined using traditional algorithms such as RRT and RRT*. Sharp turn points, bends, and straight line segments are identified, and the path is pruned and smoothened using a Bezier curve-based technique while prioritising straight path trajectories over curves for better stability. The path for each robot in the system is determined through inverse kinematics computations. The proposed planner is evaluated across diverse environmental configurations with obstacles of varying sizes to validate its performance in real-world scenarios. Achieving 98% task success and 94% obstacle avoidance, the proposed planner yields trajectories that are 25% smoother and 15% more efficient compared to standard approaches. A planning latency of 94.5 ms and replanning frequency &gt; 10 Hz demonstrates real-time, scalable coordination for MCDPM systems.</p>