Lucia Donatelli

Georgetown University, Saarland University

Papers

Total Citations

H-Index

About

Lucia Donatelli is a leading researcher at the intersection of computational semantics, natural language understanding, and human-robot interaction. Her work centers on making Abstract Meaning Representation (AMR) a viable semantic framework for situated dialogue systems, particularly in high-stakes domains like search and rescue. In her most-cited work (17 citations), she proposed 36 augmented AMRs that capture speech acts and contextual grounding, directly addressing the gap between abstract semantic graphs and the demands of real-world robot control. She further developed graph-to-graph transformation techniques to map these rich semantic representations into constrained robot action specifications, enabling more natural two-way communication. Donatelli also led the creation of the SCOUT corpus (2024), a multi-modal collection of human-robot dialogue from Wizard-of-Oz experiments in collaborative exploration. This resource is foundational for training and evaluating NLU systems that must handle situated, multi-modal instructions. Her contributions are critical for advancing robots that can understand not just what is said, but what is meant in context—a key challenge for autonomous systems operating in dynamic, unpredictable environments.