BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Synthesizing Diverse Policies for Multi-Agent Coordination - Prof.
  Amanda Prorok (Cambridge)
DTSTART:20260507T130000Z
DTEND:20260507T143000Z
UID:TALK243376@talks.cam.ac.uk
CONTACT:Dr Luca Cocconi
DESCRIPTION:How can we effectively orchestrate large teams of robots and t
 ranslate high-level goals into the nuanced local policies that guide indiv
 idual robot behavior? Machine learning has revolutionized how we address t
 hese challenges\, enabling the automatic synthesis of agent policies direc
 tly from task objectives. In this presentation\, I will first describe how
  we use data-driven approaches to learn interaction strategies that foster
  coordination and cooperation within robot teams. I will then discuss meth
 ods for learning heterogeneous policies\, where robots adopt different rol
 es\, and explain how this approach overcomes limitations inherent in tradi
 tional homogeneous models that force all robots to behave identically. Und
 erpinning this work is a measure of 'System Neural Diversity\,' a tool tha
 t allows us to quantify the degree of behavioral heterogeneity within mult
 i-agent systems. I will demonstrate how this metric enables precise contro
 l over diversity in multi-robot tasks\, leading to significant improvement
 s in performance and efficiency\, and unlocking the potential for novel an
 d often surprising collective behaviors.
LOCATION:Seminar Room 3\, RDC
END:VEVENT
END:VCALENDAR