Adam Nichamoff
Research Mentor(s): Vasileios Tzoumas
Mentor Department: Aerospace
Authors: Adam Nichamoff, Vasileios Tzoumas
Session: Session 7 (4:00pm – 4: 50pm)
Presentation Type: Poster 65
Abstract
We attempt to develop a rigorous formulation for decentralized coordination paradigm, in which autonomous heterogeneous robotic agents coordinate and communicate amongst themselves to ensure efficient task-allocation and efficient task-completion. We often depend on heterogeneous systems of robots, to complete large-scale and/or information-intensive tasks, such as surveying an unknown area, conducting a search and rescue operation, or tracking a target. Furthermore, in practical settings, we are often resource constrained (e.g. limited by time, manpower, or money) and resource/time-efficient task-completion is paramount, to ensure the feasibility of a given task. Our paradigm calls upon a variety of mathematical and computational tools to ensure efficiency, primarily mathematical optimization tools. The main focus of our research is on the utility of submodularity, game theory, and control theory, to help formulate an efficient coordination paradigm for heterogeneous multi-robot networks. Furthermore, our paradigm will also make use of online optimization and stochastic techniques to allow for multi-robot networks to adapt to changing conditions, of the robotic agents, environmental conditions, tasks, and priorities. The online, resource-constrained nature of our coordination paradigm are key distinguishing characteristics of our coordination paradigm. We pay special attention to the energy and time efficiency of computation and communication within multi-robot networks. To this end, we opt for decentralized computations and communications, with only local information, instead of being reliant upon a central computation unit or the relaying of information from all robots to all other robots across their communication network.
