A Survey on Intermediation Architectures for Underwater Robotics

Abstract

Currently, there is a plethora of solutions regarding interconnectivity and interoperability for networked robots so that they will fulfill their purposes in a coordinated manner. In addition to that, middleware architectures are becoming increasingly popular due to the advantages that they are capable of guaranteeing (hardware abstraction, information homogenization, easy access for the applications above, etc.). However, there are still scarce contributions regarding the global state of the art in intermediation architectures for underwater robotics. As far as the area of robotics is concerned, this is a major issue that must be tackled in order to get a holistic view of the existing proposals. This challenge is addressed in this paper by studying the most compelling pieces of work for this kind of software development in the current literature. The studied works have been assessed according to their most prominent features and capabilities. Furthermore, by studying the individual pieces of work and classifying them several common weaknesses have been revealed and are highlighted. This provides a starting ground for the development of a middleware architecture for underwater robotics capable of dealing with these issues.

Towards a Hybrid Approach to Context Reasoning for Underwater Robots

“…Existing context-aware frameworks have been proposed mainly for Smart Homes, Smart Cities, Ambient Assisted Living, Smart Grids, etc. Introducing the technique of context-aware framework into the field of underwater robots is under-researched [12]. Only a few attempts, including Pandora [22], Trident [23], and CoCoRo [24], have been presented for delivering context awareness in underwater robots.…”

“…Ontologies, which are considered the most promising modeling technique in terms of expressiveness and interoperability [10], have been adopted in the context-aware framework to formally represent heterogeneous contexts that are exchanged between vehicles, such as the battery level of vehicles, capabilities of vehicles, turbidity, position, speed, salinity, and wind direction [11]. It is worth noting that operators or vehicles are more willing to use high-level context information instead of raw context to conceive and offer context-aware services [12]. For instance, a piece of high-level context, vehicle A might collide with vehicle B soon, is more meaningful for operators to make decisions than the basic context information vehicle A is out of trajectory and heading in the direction of vehicle B.…”

“…In particular, the harsh nature of the underwater environment and the limitations of sensors attached to vehicles impose more challenges in making effective context reasoning in underwater robots. In the underwater environment, the majority of contexts, namely sensory data, are prone to be uncertain [12]. Uncertainties, as an inherent characteristic of context, should be handled in the reasoning phase.…”

SWARMs Ontology: A Common Information Model for the Cooperation of Underwater Robots

“…In order to achieve this ambitious goal, the SWARMs project must tackle a set of challenges [5], e.g., defiant maritime and underwater communication and data heterogeneity.…”

“…The context reasoner employs a hybrid context reasoning method [9], which includes ontological, rule-based, and Multi-Entity Bayesian Network (MEBN) reasoning. It is worth noting that information that is obtained by sensors or other sensing instruments is prone to be uncertain in the underwater environment [5] due to partial views, imperfect instruments, challenging communications, data loss, etc. Thus, uncertainty, as an inherent characteristic of information, also needs to be formally annotated in the SWARMs ontology for the sake of completeness in knowledge representation and also to provide support for uncertainty reasoning defined in the context reasoner component.…”