Collaborative Autonomy, GNNs, and Spectrum

Autonomous multi-agent UAV and UGV systems are promising for critical applications such as disaster response, logistics and transportation, supplementing cellular networks, agricultural sensing, environmental monitoring, and security and military operations. Achieving the desired multi-agent collaborative behaviors requires joint perception and information sharing over wireless links. In this talk we look at recent progress on graph neural network (GNN) approaches for learning perception-action-control (PAC) loops. This includes learning what information to share among agents that supports a team objective. Going forward, many issues arise, including how to combine learning and model-based control, achieving scalability and resilient group behavior, and melding mobility and networking. The combination of learning, modeling, and cognitive networking is a promising path to collaborative intelligent systems. Speaker(s): Brian M. Sadler, Agenda: 6:30 PM Welcome and introduction 6:35 PM Presentation 7:20 PM Q&A 7:30 PM Conclude Virtual: https://events.vtools.ieee.org/m/380463