CAREER: Scalable and Reliable Coordination in Embodied Intelligent Networks: A Submodular Optimization and Online Learning Perspective

Embodied intelligent networks are collections of distributed autonomous agents that can sense, reason, communicate, and act, leveraging suggested commands by human operators and/or external machine-learning algorithms. Scalable and reliable coordination among such agents can benefit society in tasks that range from environmental monitoring to transportation to national defense.

But achieving scalability is challenging due to the agents' limited resources vs. their resource-demanding tasks that are often combinatorial and NP-hard. Achieving reliability is challenging due to (i) the agents' limited observability the environment, (ii) the environment's unpredictability, and (iii) the external commands' untrustworthiness, i.e., their lack of performance guarantees.

This CAREER proposal will lay the theoretical and algorithmic foundation to overcome these challenges by introducing coordination and online-learning capabilities that enable multi-agent networks to (i) self-configure their communication topology to balance the trade-off of scalability vs. coordination performance, (ii) adapt online to unpredictable environments, and (iii) reap the benefits of external commands, managing the risks of erroneous commands. The proposed combinatorial optimization approach will be transformative by characterizing the trade-off between scalability and coordination performance via submodularity theory, and by contributing online-learning coordination algorithms that can balance the trade-off by tuning the degree of decentralized coordination among the agents, even in unpredictable and untrustworthy environments.

The proposed research efforts will be evaluated in information gathering tasks via both physics-based simulations and field experiments. The research outcomes will provide practical methods for the distributed intelligence of critical infrastructure networks of the future such as multi-robot networks with humans in the loop, air-land-sea connected autonomous vehicles, and ubiquitously deployed sensors.

Applications range from information gathering for environmental monitoring, disaster response, and surveillance to motion coordination and monitoring for traffic control in smart cities.

The research efforts will also be integrated with educational efforts to enable a diverse future workforce that can design, deploy, and interact with embodied intelligent networks. Specifically, the planned educational efforts will engage students from the middle-school level up to the undergraduate level. The efforts will culminate in hands on research experiences with multi-quadrotor systems equipped with the proposed decision-making and decentralized-communication capabilities.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.