ConceptBot
Knowledge-graph-grounded commonsense reasoning for LLM-based robot task planning.
Research / AI & Robotics
Alessandro Leanza
PhD Candidate in AI & Robotics at IDSIA USI-SUPSI
Embodied AI · Multimodal Robot Learning · Generative AI · LLM/RAG
I develop AI systems that connect reasoning, perception, and physical interaction, from knowledge-grounded LLM planning to multimodal robot learning and adaptive human–robot systems.
Selected Projects
All projects
Context-Aware Exoskeleton Assistance
Multimodal perception and adaptive control combining gaze, vision, and physical interaction.
Graph-Based RAG for Robot Planning
Local, provenance-aware retrieval and knowledge graphs for grounded robot task decomposition.
Learning Compliance with Diffusion Policies
Work in progress: learning motion and compliance schedules from demonstrations, with force-aware diffusion policies and velocity-level admittance control.
Research Focus
Embodied AI
Generative AI & Reasoning
Multimodal Robot Learning
Selected Publications
ConceptBot: Knowledge-Graph–Grounded Commonsense for Task Decomposition in LLM Robot Planning
Alessandro Leanza, Angelo Moroncelli, Giuseppe Vizzari, Francesco Braghin, Loris Roveda, Blerina Spahiu
Accepted in 2026 for publication in Biomimetic Intelligence and Robotics (BIROB).
2026 · final journal citation pending
About
Research & engineering
I am a PhD candidate in Informatics at IDSIA USI–SUPSI in Lugano, Switzerland. My research combines generative and multimodal AI with robot learning, from knowledge-grounded LLM planning to compliant manipulation and adaptive human–robot interaction.
More about Alessandro