I’m thrilled to share that at IEEE ICRA 2026 in Vienna, our research on natural language-driven robot skill selection and adaptation received two Next-Gen Spotlight Awards — one for each of our frameworks, CLASP and IROSA.
The awards were presented at the workshop Bridging the Gap between Robot Learning and Human-Robot Interaction, where six selected awardees gave spotlight presentations on work spanning shared control, language-driven robot skills, tactile human-intent recognition, and physical human-robot interaction.
CLASP — Skill Selection and Composition Link to heading
CLASP combines Vision-Language Models with Task-Parameterized Kernelized Movement Primitives for data-efficient, language-driven robot skill selection, composition, and active acquisition — requiring only 2–5 demonstrations per skill. [Paper on ArXiv]
IROSA — Interactive Skill Adaptation Link to heading
IROSA enables open-vocabulary robot skill adaptation through natural language using a tool-based architecture, allowing speed adjustments, trajectory corrections, and obstacle avoidance — zero-shot, without any fine-tuning. [Paper on ArXiv]
Acknowledgments Link to heading
A huge thank you to the organizers for the recognition, and to my co-authors Valentin Gieraths, Samuel Bustamante, Tai Mai, João Silvério, Thomas Eiband, Freek Stulp, and Alin Albu-Schäffer for making this work possible.