This paper presents and experimentally validates a vision-guided, synchronized cooperative manipulation framework for two DJI RoboMaster EP mobile manipulators performing a cylinder pick-and-place task in a warehouse-relevant setting. The main novelty of the proposed system lies in its event-driven synchronization mechanism for dual-robot cooperative manipulation, which enables coordinated grasping and transport without direct robot-to-robot wireless messaging, depth sensors, external motion-capture infrastructure, or pre-built 3D maps. The complete manipulation pipeline includes visual lateral alignment, visual forward approach, grasp execution, synchronized lifting, cooperative transport, and coordinated release. Visual guidance is achieved using color-based HSV detection and proportional control, providing a lightweight perception and control solution for cooperative object handling. The framework is evaluated through both fixed-position repeatability and position generalization studies. In a five-trial repeatability study, the system achieved a 100% task success rate with a mean completion time of 65.26 ± 1.65 s and a coefficient of variation of 2.5%. In a six-configuration generalization study, the system again achieved 100% success, with task completion times ranging from 50.8 s to 86.5 s. Across all eleven trials, the proposed synchronization scheme maintained safe and correct task progression across all trials. In addition, grip synchronization latency increased with asymmetry in the robots’ approach distances to the object, with symmetric configurations producing latencies below 6 s and asymmetric configurations exceeding 22 s. These results suggest that event-driven synchronization can provide a practical and lightweight coordination mechanism for cooperative dual-robot manipulation under controlled laboratory conditions with minimal sensing and communication requirements.