CURRENT PROJECTS

Top-Down Synthesis of Robot Ensemble Behavior

The focus of this project is to develop scalable control and coordination strategies for robot ensembles. The principal objective of this work is to combine the power of existing multi-robot and robot swarm approaches and to create novel techniques for the modeling, analysis, and synthesis of robot ensemble behaviors. This work leverages existing techniques from chemical reaction network theory (CRNT) to address the controller synthesis problem for distributed robot ensembles. CRNT is especially well suited to model cooperative ensemble behaviors because it has the ability to explicitly model physical interactions between various elements and to describe their collective dynamics. Central to this approach will be the development of a robot reaction network theory (RRNT) that retains all the essential features of CRNT, and incorporates additional elements, such as task execution times, precedence constraints, and nonuniformity in spatial distribution, to handle conditions that are common in robotic applications. These efforts will also include the development of a robotic reaction network testbed to evaluate the soundness of the proposed models and methodologies.

Social Selection within Forked Fungus Beetles

The objective is to study the Forked Fungus Beetles to explore complex systems operating in nature to address critical gaps in our understanding of social networks and to inform the design of scalable coordination strategies for large heterogeneous teams of mobile robots. These insects interact in locally substructured social networks, respond dynamically by changing behavioral states based on interactions with other individuals, exist in unpredictable and changing environmental contexts, and most importantly, exhibit emergent patterns of spatial and behavioral distribution at the population level. To reveal the key local interaction patterns that lead to desirable group outcomes, we need to address the technical challenges involved in the development of a network of smart camera nodes that can be used to track every member of an entire population of beetles as they move across the full range of their habitat. Such a system will provide measurements of the beetles’ activities with an unprecedented level of spatial and temporal resolution. This data will allow us to determine how specific patterns of interaction lead to social network structures within natural populations and how one can translate these interactions into the design of behavioral strategies for mobile robot teams that lead to desired group-level outcomes. In essence, how emergence can be engineered through the controlled modification of individual robot behaviors.