The overactuated unmanned aerial vehicles (UAVs) platform, composed of multiple UAVs, enhances the payload capacity and flexibility of aerial platforms effectively. However, due to the potential motion conflicts among UAVs, accurate tracking of trajectory remains challenging. The connection mechanism is designed to link multiple UAVs together to form the overactuated UAVs platform. To improve the trajectory tracking accuracy of the overactuated UAVs platform, a decision-making framework based on reinforcement learning is proposed. Besides, an automatic weighting of reward method is introduced to improve the fitness of the proposed framework on overactuated UAVs platform. By automatically generating the weights of reward, this method effectively reduces conflicts among UAVs and enhances the coordination of the platform. Finally, simulations are conducted to track the desired trajectory of the overactuated UAVs platform in two scenarios. The simulation results confirm the effectiveness of the proposed decision-making framework in reducing position errors.