ISLa - Intelligent Systems Lab

BEHAVE - Learning Safe Behaviours for human-robot cooperation

The BEHAVE project develops new AI methodologies to assess and certify the quality of decisions for intelligent robotic systems. BEHAVE operates within Spoke 5 in cooperation with the other Spoke partners and will focus on planning and cooperation between humans and robots. In particular, the BEHAVE project assesses issues related to the formal verification of Deep Neural Networks trained using Deep Reinforcement Learning approaches, in order to certify the behaviors of robots interacting with humans.

Starting date: December 2, 2024

Duration: 10 months

Deparments: Dept. of Computer Science, University of Verona

Manager: Prof. Alessandro Farinelli

Sponsors: MUR - Ministero dell’Università e della Ricerca (assigned and managed by the department)

Participants

  • Manuele Bicego – Associate Professor
  • Alberto Castellini – Associate Professor
  • Ferdinando Cicalese – Full Professor
  • Alessandro Farinelli – Full Professor
  • Isabella Mastroeni – Associate Professor
  • Daniele Meli – Temporary Assistant Professor

Research Group Involved

  • Luca Marzari – PhD Candidate
  • Celeste Veronese – PhD Student

BEHAVE project image

  • [AAMAS 2025] Bonanni, L., Meli, D., Castellini, A., & Farinelli, A. (2025). Monte Carlo Tree Search with Velocity Obstacles for safe and efficient motion planning in dynamic environments. In Proceedings of the 24th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2025), pages 371-380, Detroit, MI, USA, ISBN: 9798400714269. https://dl.acm.org/doi/10.5555/3709347.3743551.

  • [IAS 2025] Veronese, C., Meli, D., Iocchi, L., & Farinelli, A. (2025). Symbolic Exploration with Partial Models in Discrete Reinforcement Learning. Accepted for presentation at the 19th International Conference on Intelligent Autonomous Systems (IAS-19), Genoa, Italy, June 30 – July 3, 2025, to be published in the conference proceedings.

  • [RLC 2025] Zorzi, E., Castellini, A., Bakopoulos, L., Chalkiadakis, G., & Farinelli, A. (2025). Seldonian Reinforcement Learning for Ad Hoc Teamwork. Accepted for presentation at the Reinforcement Learning Conference (RLC), Edmonton, Canada, August 5 to August 9, 2025, to be published in the Reinforcement Learning Journal. https://arxiv.org/abs/2503.03885.