Internet is a dynamic and unstructured network with multi-domain and multi-administration. ROS and ROS2(DDS) are not designed for the unstructured network, hence won’t work in real-world deployment. We need an effective mechanism to unblock ROS traffic and orchestrate internet resources for autonomous robots. The mechanism could accelerate the transformation from existing cloud robotics into the future of 5G/6G enabled connected robotics. This tutorial presents a ROS extension called Connected Robotics Platform (CROP) to extend the ROS from only working in structured lab environments into fully scalable at real-world CONNECT+ settings. CROP is an open source, cloud native implementation of Network Function Virtualisation (NFV) which optimised specifically for robot development. Existing ROS applications enhanced by CROP can be deployed natively into large-scale distributed systems without modification, at the same time maintain the necessary performance and security. Furthermore, a radio-aware semantic mapping framework is introduced. By combining radio signal quality information with semantic mapping techniques, the framework creates a comprehensive understanding of the network environment, which enhances the decision-making of the autonomous robots. The development is a step forward to 5G/6G enabled connected robotics on supporting robot deployment in real-world settings at the same time optimising behaviours of robots according to digital infrastructures.