This paper presents an innovative approach to multi-robot path planning within Robotic Compact Storage and Retrieval Systems (RCSRS) by integrating Mixed-Integer Linear Programming (MILP) with the A* algorithm. Facing the challenge of optimizing paths in densely populated, dynamic environments, we propose a solution that combines the comprehensive optimization capabilities of MILP with the heuristic-driven efficiency of the A* algorithm. Through detailed mathematical modeling and algorithmic strategies, our approach addresses critical issues such as collision avoidance and computational efficiency. Simulated experiments in a typical RCSRS environment demonstrate the effectiveness of our method, showing significant improvements in pathfinding optimality and computational time. This study contributes to the fields of logistics, warehousing automation, and robotics by offering a scalable and efficient solution for the complex problem of multi-robot path planning, paving the way for future research in the optimization and practical application of autonomous robotic systems in compact storage spaces.