Chair: Dimitra Panagou

Zoom passcode: hri

Abstract:
This dissertation investigates how humans can effectively supervise autonomous systems—particularly multi-robot teams—across a range of control configurations and interface designs. While much of robotics research focuses on either fully manual control or fully autonomous operation, real-world applications increasingly demand a middle ground, where perception, decision-making, and control are distributed between human and machine. Through four case studies, this work examines supervisory scenarios that vary in control authority and information flow: from autonomous robot formations that adapt to uncertain human motion, to mixed reality interfaces for spatial command, to passive haptic feedback for improving takeover performance, and finally to collaborative interventions for failure recovery. Taken together, these studies contribute new tools, insights, and empirical evidence toward designing supervision not as a fallback mode, but as an active, embodied, and scalable form of human–robot interaction.