4. Juni 2026, von Theresa Pekarek Rosin

Foto: UHH Knowledge Technology

We are pleased to announce that the paper "HEXAR: A Hierarchical Explainability Architecture for Robots", authored by several DCs and PIs in the TRAIL project, has just won the Best Paper in Human-Robot Interaction award at ICRA 2026. The paper proposes a novel architecture for explaining robot behaviour, leveraging a set of specialised component explainers (designed to explain particular robot modules) orchestrated by an explainer selector, which chooses the most suitable explainers to answer a user question given relevant context. Results across 180 samples over 20 scenarios in a domestic use case show that HEXAR provides accurate explanations which minimise false information.

Congratulations to the authors:

Tamlin Love (DC - IRI CSIC-UPC), Pradip Pramanick (DC - University of Naples Federico II), Ferran Gebellí Guinjoan (DC - PAL Robotics), Antonio Andriella, Guillem Alenya (PI - IRI CSIC-UPC), Anais Garrell, Raquel Ros (PI - IIIA CSIC), Silvia Rossi (PI - University of Naples Federico II).

See the paper website at: https://pradippramanick.github.io/hexar/