This paper describes a system based on a robot, called KuBo, which relies on cloud resources to extend its capabilities for human interaction and environmental sensing to provide services for independent living. The robot uses text-to-speech and speech recognition services as well as an electronic agenda and web resources to perform several tasks. Moreover it retrieves smart environmental data from a DataBase to be aware of the context. In this paper, the cloud robotics approach is used to increase the skills of a robot, endowing the system with abilities for human-robot interaction and environmental sensing. The robotic services have been defined with a focus group involving 19 elderly volunteers and the system has been tested in a real environment with a couple of elderly users for five days. The aim of the experiment was to test the technical feasibility of the proposed cloud services using quantitative tools. The technical results show a success rate of 86.2 % for the navigation task and more than 90 % for the speech capabilities. Furthermore, the robustness of the system was also confirmed by users' qualitative feedback.