In recent years, the research of cloud robots has gained widespread academic attention for their rapid development. As the robotic system and function become more and more complex, it is more difficult to implement the robotic control algorithm in the edge system accordingly, and the programming language and execution environment of multi -vendor robots are also significantly different. In order to solve the robotic control problems in cloud computing environments, including the programming diversity, the online-test complexity and the difficulty of the collaborative control for multi-robot systems, this paper proposes a new control platform architecture for cloud robots, called IAPcloud, in which the virtual controller technology is introduced to solve the collaborative control issue among heterogeneous robots or robots and devices. A hardware-agnostic communication technology of the edge controller for heterogeneous robots or devices is also presented. The proposed platform architecture has been tested in a designed cloud robot experimental system by two cases, including the image recognition and tracking case and the human-machine gobang game case. The results indicate the effectiveness of our approach for simplifying the cloud robotic programming method, shorten the development and deployment cycle, and tackling the complex collaborative control task of heterogeneous robots or devices.