Communicationless Coordination via Intent Estimation and Intent-Expressive Motion Planning in Multiagent Exploration

This project will design algorithms for communicationless coordination of multi-robot systems via intent estimation and intention-expressive motion planning. In collaboration with Geisel Software, Inc., the ASU team will explore the feasibility and design of these algorithms by building upon prior work on intent inference for dynamical systems, e.g., autonomous driving and swarm intent estimation. In particular, the ASU team will develop behavior and intent estimation/prediction algorithms that combine set-based and probabilistic model discrimination and estimation frameworks to detect a potential mismatch between intended and actual paths/trajectories of rovers. Moreover, the cooperative team will design intent-expressive (also known as lsquo;legiblersquo;) motion planning algorithms such that the intended paths/trajectories and tasks are more easily inferred without explicit communication. Furthermore, in this first phase, we will perform feasibility and test studies for the proposed communicationless coordination approach by developing a realistic simulation platform such as a Gazebo simulator and leveraging Geisel Softwarersquo;s prior experience and expertise in this area. To complement the roles of the collaborators at Geisel Software, Inc., Dr. Yong and his graduate student will assist with algorithm design and simulations.