Figure 1 - uploaded by Julio Dondo
Content may be subject to copyright.
Architecture of the Scone System-on-Chip. 

Architecture of the Scone System-on-Chip. 

Source publication
Article
Full-text available
Enabling Ambient Intelligence systems to understand the activities that are taking place in a supervised context is a rather complicated task. Moreover, this task cannot be successfully addressed while overlooking the mechanisms (common-sense knowledge and reasoning) that entitle us, as humans beings, to successfully undertake it. This work is base...

Contexts in source publication

Context 1
... these goals in mind, a hardware platform, based on the use of FPGAs (Field Programmable Gate Arrays) for fast reasoning under the Scone framework has been proposed. In Figure 1, a high level picture of the proposed reasoning hardware platform (from now on RHP) is depicted. The system is composed of the following elements: ...
Context 2
... a dynamic point of view, the proposed platform can freely grow avoiding an oversized initial configuration. If needed, a new SN can be instantiated in a free partial reconfiguration area (see white dotted box in Figure 1) in order to expand, for example, the semantic network with a new branch defining a subclass of the thing supernode. New entries can be added to the existing SNs through writing operations performed on local memories. ...
Context 3
... mentioned in previous sections, the marker-passing algorithm implemented by Scone is one of its main strengths. Briefly, the marker-passing mechanism is mainly devoted to extract the implicit knowledge that it is contained in the semantic network by means of propagating pieces of information through the complex web formed by nodes and links. Thus, three main challenges arise when facing the hardware implementation of such a kind of knowledge-base systems: (a) to simplify the representation and storage of the semantic network information; (b) to optimize the memory organization for a fast and efficient implementation of the marker-passing algorithm; and (c) to provide a scalable and distributed architecture that simplifies the task of adding new knowledge while at the same time being capable of performing parallel searches. Bearing these goals in mind, a hardware platform, based on the use of FPGAs ( Field Programmable Gate Arrays ) for fast reasoning under the Scone framework has been proposed. In Figure 1, a high level picture of the proposed reasoning hardware platform (from now on RHP) is depicted. The system is composed of the following elements: • A Microblaze processor running at a frequency of 125 MHz. There is no other operating system running in the platform but a little software layer ( Xilkernel ) implementing basic ...
Context 4
... device chosen for the implementation of this prototype is a XUPV5 board ( from Xilinx. It is based on a Virtex5 LX110T chip, equivalent to four million logic gates with run-time partial reconfiguration capability. Our reasoning hardware platform takes advantage of this dynamic feature provided by the FPGA in order to adapt itself to unforeseen scenarios while beings scalable. For example, the number of SNs instantiated in a given moment depends on the size of the semantic tree capturing the information for the application under execution. From a static point of view, unused SN instances are unnecessarily wasting power consumption. From a dynamic point of view, the proposed platform can freely grow avoiding an oversized initial configuration. If needed, a new SN can be instantiated in a free partial reconfiguration area (see white dotted box in Figure 1) in order to expand, for example, the semantic network with a new branch defining a subclass of the thing supernode. New entries can be added to the existing SNs through writing operations performed on local memories. The work proposed here resorts to the infrastructure for dynamic reconfiguration management for Xilinx platforms developed by Dondo et al . [22]. Another scenario that is worth being mentioned refers to the fact that it is possible to face a run-time rearrangement of the semantic tree data among the SNs. For optimization purposes (see Section 3.5) it would be necessary to split one branch of the semantic tree (actually stored in one SN) and consequently more SNs will be instantiated to hold the resultant subtrees. This section is aimed at presenting an overview of how the RHP works at system level, before getting into the implementation details of the individual Semantic Nodes. From the user’s perspective, the presence of the RHP is totally transparent since it is remotely used through the standard Scone ...

Citations

... For the same reason, the restrictions of this type of inference still apply. For instance, other types of reasoning [11] may not be applied and it would be interesting to take into account the context uncertainty and vagueness [12] so as to better model the environment. ...
Article
Full-text available
Healthcare environments, as many other real world environments, present many changing and unpredictable situations. In order to use a social robot in such an environment, the robot has to be prepared to deal with all the changing situations. This paper presents a robot self-configuration approach to overcome suitably the commented problems. The approach is based on the integration of a semantic framework, where a reasoner can take decisions about the configuration of robot services and resources. An ontology has been designed to model the robot and the relevant context information. Besides rules are used to encode human knowledge and serve as policies for the reasoner. The approach has been successfully implemented in a mobile robot, which showed to be more capable of solving situations not pre-designed.