Towards Cloud-Based Architectures for Robotic Applications Provisioning

Robotic applications are widely used in various domains, such as healthcare, precision agriculture, and disaster management. However, provisioning them in a cost-efficient manner remains an uphill task. Cloud computing is changing the way applications are provisioned. It is a new paradigm with three key facets: Infrastructure as a Service (IaaS), Platform as a Service (PaaS) and Software as a Service (SaaS). Access anytime, anywhere, rapid application development through third party application re-use, and efficient use of resources are among cloud computing’s expected benefits. All these expected benefits can potentially lead to cost efficiency in robotic applications provisioning. This thesis defines a set of requirements for cost-efficient robotic application provisioning. It starts by reviewing the state of the art. A key issue from business perspective is how robotic application and third party application clouds could interact for cost efficient robotic applications provisioning purpose. This thesis proposes a business model that can potentially solve the issue. It is made up of four categories of actors: the end-users, the robotic cloud providers, the third party cloud providers and the cloud interaction framework provider. It is actually the cloud interaction framework provider that enables the interactions between robotic application clouds and third party application clouds. We also propose a peer to peer (P2P) overlay network as the basis of the architecture of the interaction framework provider role. It is a P2P overlay network in which each cloud provider (i.e. robotic cloud and third party cloud providers) is represented by a node for publication, discovery, activation, and execution of applications purposes. As a proof of concept, we implemented the P2P overlay based architecture including the P2P overlay network with its nodes. We have also implemented a robotic application (i.e., wildfire suppression) and the third party applications it re-uses (i.e., fire detection and supply management) with robotic application and third party applications residing in separate clouds. In addition, performance measurements have also been made