Efficient knowledge sharing, computation load minimization, and collision-free movement are very important issues in the field of multi-robot automation. Several cloud robot architectures have been investigated to fulfill these requirements. However, the performance of the cloud-robot architectures created to date are suboptimal due to the lack of efficient data management for multi-robotic systems. With this point in mind, this paper proposes an efficient cloud multi-robot framework with cloud database model for mobile robot applications to facilitate multi-robot management, communication, and resource sharing. In this proposed architecture, the cloud framework is comprised with cloud data analysis, cloud database management, and cloud service management. The data analysis serves different data processing and decision-making tasks for generating the next robot action based on robot sensors’ data with the help of a data access components layer. A multistage cloud database model distributes, stores, and accesses different categories of data related to robot sensors and environments. And cloud service facilitates multi-robot management, communication, and resource sharing in the cloud framework. Additionally, as a use case, a cloud-based convolutional neural network (CNN) model is introduced for learning and recognizing robot application data. The obtained results of our tests indicate that the proposed cloud-robot architecture provides efficient computation power, communications, and knowledge sharing for managing multi-mobile robot systems.