The massive storage and parallel computation that the Cloud offers via an Internet connection are enabling intelligent systems to tackle new problems, and opening up new application domains. In the area of Automation, these newly available possibilities allow storing and leveraging massive amounts of data, explicitly reasoning about uncertainty by considering multiple hypotheses about the state of the world, and operating in more complex environments than previously possible. Cloud Robotics and Automation explores how the newly emerged cloud paradigm can enable robots and automated systems to operate far beyond their traditional roles and in new domains.
The goal of this session is 1) to bring together the latest research on the field and 2) provide a forum to discuss the challenges and opportunities associated with the cloud technologies and its use in automation applications. Some of the most relevant topics of the field are
- Leveraging existing cloud technologies and resources. E.g.: To what extent can existing IaaS, PaaS, and SaaS resources be used or adapted for robotics and automation? What algorithmic, technical, or conceptual advances are needed to allow robots to use the powerful computational, storage, and communications infrastructure of modern data centers?
- Leveraging synergies with the Internet of Things. E.g.: How can robots and automation systems get networked into the IoT? What advances are required to complement the IoT’s sensor technologies with a physical layer for actuation?
- Online knowledge bases and storage technologies. E.g.: How will/should online knowledge bases grow? What are the processes leading to creation of a substantial knowledge base useful in real applications?
- Data mining. E.g.: How can we extract patterns or relationships between variables from large data streams coming from robotic or automation platforms? How can the automated systems benefit from the recognition of such patterns?
- Knowledge discovery, representation, reuse, and interoperability. E.g.: How can we extract abstract concepts from sensor/actuator data? How can a robot decide which knowledge (e.g., map or skill) to reuse in a new situation? How the data gathered from one particular system can be used by very different ones? How can we ground abstract knowledge or complex tasks on low-level commands in specific systems and robots?
- Leveraging existing big data. E.g.: How can robots and automation systems exploit the vast amounts of weakly-supervised data that general-purpose repositories (e.g., Google or Youtube) offer?
- Platform design. E.g.: What architectures provide the optimal trade- offs between content aggregation and caching vs. accessibility and scalability vs. response time for robotics and automation applications? What are useful metrics and optimal trade-offs between on-board computation and the use of cloud services?
The session builds on the success of an IEEE T-ASE Special Issue on Cloud Robotics and Automation (to appear in April 2015) and several previous workshops (among others, the recent Cloud Robotics Workshop at IROS 2013 and the Cloud Manufacturing and Automation workshop at CASE 2013); all of them co-organized by the proposers of this session. In this occasion, and differently from previous workshops, we target both the robotics and automation communities. We believe that the research problems are similar in both applications, and hence researchers from both sides can benefit from each other’s views.