This paper introduces a fully integrated task and motion planning framework for robotic manipulation in dynamic and uncertain environments, particularly in multi-arm manipulator systems where reachability constraints change over time. Existing approaches, including probabilistic road maps and traditional task-and-motion planning (TAMP) frameworks, struggle with high-dimensionality, infeasible actions, and limited adaptability. To address these challenges, we propose a novel planning system that seamlessly integrates high-level symbolic task planning with low-level motion execution via a Dynamic Shared-Space Graph (D-SSG) and a Planning Domain Definition Language (PDDL)-based interface layer. The key contribution of this work is a real-time re-planning mechanism that dynamically adjusts to task failures, geometric constraints, and environmental variations, ensuring robust execution. Extensive simulations in ROS and Gazebo demonstrate a measurable reduction in planning time and a significant increase in task success rates under cluttered and uncertain conditions. These results establish the proposed framework as a promising solution for advancing multi-robot coordination in industrial and service automation applications.