Rafael Barea

Abstraction, Synthesis, and Integration of Information for Human-Robot Teams (ABSYNTHE) - UAH

Research areas:
Status: Finished
Type Project: Public_National
Project leaders: Manuel Ocaña
Collaborators: Elena López; Ricardo García; Pedro Revenga; Ramón F. Flores; Noelia Hernández; Fernando Herranz; Ángel Llamazares; Eduardo Molinos; Jorge Pozuelo
Funded by: MINISTERIO DE ECONOMÍA Y COMPETITIVIDAD
Funding: 58443.00 EUR
Proposed start date: 2012-01-01
Proposed end date: 2014-12-31
Description:

Robots are increasingly being deployed in teams that support a variety of human activities, including planning and decision-making. Beyond supporting human endeavors, these teams often include humans as active components that sense, interact, and collaborate with their automated counterparts. These robotic teams are characterized by their ability to acquire vast amounts of data that must be properly processed, analyzed, interpreted, and fused to ensure its right functioning, in terms of coordination, negotiation, distribution, and cooperation. 

However, one of the biggest problems for the implementation of human robots teams is that the existing information-processing systems can not emulate important cognitive capabilities of human beings due, to a large extent, to the lack of approaches to generate and effectively abstract salient semantic aspects of information coming from different sources. On the other hand, the competency of humans to collaborate in the solution of complex problems, while exchanging information that facilitates the understanding of system behavior, may be explained by their use of cognitive tools that produce summarizations and descriptions of complex objects, events, and relations in terms that are easy to comprehend by other human beings.

This ability to abstract and summarize knowledge at various levels of description is an essential trait of cognitive intelligent processes that is not, at present, adequately emulated by information processing systems. Furthermore, lack of tools that represent computational objects in terms that facilitate human understanding is a major barrier to the effective collaboration of human and robots in mixed human/machine environments, in complex tasks such as rescue, scouting, or transportation.

In this context, the main aim of this project is the development of concepts, tools, and approaches for the collaborative production and application of high-level semantic descriptions of computational objects with a view to facilitate the joint intelligent processing and exploitation of knowledge by mixed teams of human and robots. For this purpose we plan to employ a variety of soft-computing techniques, notably fuzzy logic and evolutionary computation, in combination with other approaches, thus leading to significant advances in the treatment of the problems and issues described before. More generally, we aim to lay the conceptual bases for the multilevel, semantic description of complex objects, processes, and plans. It is important to note that, while striving to establish the bases for a high-level language for communication between members of a human-robot team, our approach is not based on either the emulation of human communication capabilities nor on any other considerations stemming from a purely linguistic perspective. Our focus, rather, is on the application of knowledge extraction and data mining techniques to abstract, from complex data objects, information elements essential to the safe and effective operation of the team.