Taxonomies of Interconnected Systems: Asymmetric Interactions in Distributed Robotics

Date: 
Tuesday, 26 May, 2015
Place: 
Seattle, USA
Hosting conference: 
2015 IEEE ICRA
Organizers: 
Ryan Williams
Andrea Gasparri
Gaurav Sukhatme
Abstract: 
Interconnected robotic systems have become the recent focus of intense investigation, particularly in the context of autonomous collaboration (such as in multi-robot or sensor systems), affording fundamental advantages in adaptability, scalability, and efficiency compared to classical single-agent solutions. As recent work has demonstrated, investigations are far-reaching across various disciplines, ranging from sampling, tracking, and coverage, mobility and topology control, to general agent agreement problems. As a matter of fact, the study of interconnected systems is remarkably complex and highly susceptible to fragmentation especially due to the diversity of the research communities involved, ranging from computer science to automation. Both a high level view of the fundamental topics that drive interconnected systems, and a fine-grained understanding of each topic is required to truly make progress in the field, and to provide an accessible starting point to new research. An effective approach to attain such goals would be to construct a taxonomy of interconnected systems. To this end, we plan to organize a series of workshops, each addressing specific subareas at the forefront of interconnected system research. For each of these research topics, our goal is to identify those properties that underlie crucial, and yet common, aspects of theory and application. We believe that such a taxonomic approach may lead towards an understanding of the current open problems in each subarea, the relationships between subareas, ultimately yielding a roadmap for new researchers connecting theory and application. This proposal represents a further step towards building such a taxonomy after the initial workshop held at IROS14 (http://asimov.usc.edu/~rkwillia/ws/iros14/), which focused on the implications of topology in robotics. In this workshop, we will focus on the common assumption of undirectedness or symmetry in multi-robot systems. Undirectedness alleviates many issues in system analyses and allows for sound and accessible theoretical conclusions. However, the undirected assumption lacks the power to yield insight into realistic systems, where asymmetries in communication, mobility, and sensing are far too often present. Further, when we consider issues such as unforeseen failures, asymmetries in the environments where sensing occurs, adversarial or competitive settings, etc., the necessity of eliminating the assumption of undirectedness becomes clear. In the context of a taxonomy, directedness also represents a natural classification of interconnected networks, and an important topic in the series. Therefore, we hope that such a taxonomic view will aid in capturing the asymmetric necessities of real-world applications, moving us towards increasingly realistic multirobot systems. As our long-term vision for the series is to truly connect workshop topics, we will incorporate notions of asymmetry in topological control methodologies, to relate to the previous IROS workshop. Finally, we aim to build the series across the robotics and control communities, hopefully establishing a bridge between novelties in multi-agent theory and the future requirements of multi-robot applications.