Publications

Check my Google scholar for the updated and complete list.

• G. Aru*, E. Piccoli*, F. Trotti*, A. Farinelli, D. Bacciu. Uncertainty-Gated Safety for Out-of-Distribution Mapless Navigation. Under review. (* Equal contribution.)
  • Studies safety for mapless navigation under out-of-distribution conditions, focusing on uncertainty-aware mechanisms for more reliable decision making.
  • Paper link: Under Review

• E. Piccoli, M. Li, G. Carfi, V. Lomonaco, D. Bacciu. Combining Pre-Trained Models for Enhanced Feature Representation in Reinforcement Learning. IBRL workshop, RLC 2025 and CoLLAs 2025.
  • Proposes Weight Sharing Attention, an architecture for combining embeddings from multiple pre-trained models into richer state representations for reinforcement learning agents.
  • Paper link: https://proceedings.mlr.press/v330/piccoli26a.html
• A. Capurso*, E. Piccoli*, D. Bacciu. FAST: Similarity-based Knowledge Transfer for Efficient Policy Learning. IEEE CoG 2025. (* Equal contribution.)
  • Introduces FAST, a similarity-based transfer framework that uses visual frames and textual descriptions to select useful source policies and reduce training steps on new racing tasks.
  • Paper link: https://ieeexplore.ieee.org/document/11114355
• J. Bell, L. Quarantiello, E. N. Coleman, L. Li, M. Li, M. Madeddu, E. Piccoli, V. Lomonaco. The Future of Continual Learning in the Era of Foundation Models: Three Key Directions. TCAI workshop, HHAI 2025.
  • Argues that continual learning remains central for foundation models through continual pre-training, continual fine-tuning, and continual compositionality.
  • Paper link: https://ceur-ws.org/Vol-4074/paper10-13.pdf
• L. Quarantiello*, E. Piccoli*, et al. A Compositional Paradigm for Foundation Models: Towards Smarter Robotic Agents. I-RIM 3D 2025. (* Equal contribution.)
  • Explores how continual learning and compositionality can make foundation-model-based robotic agents more flexible, efficient, and adaptable without full retraining.
  • Paper link: https://arxiv.org/abs/2510.18608

• M. Li*, E. Piccoli*, V. Lomonaco, D. Bacciu. I Know How: Combining Prior Policies to Solve New Tasks. IEEE CoG 2024. (* Equal contribution.)
  • Presents the I Know How framework, which reuses and combines prior policies so reinforcement learning agents can adapt to new tasks more efficiently.
  • Paper link: https://ieeexplore.ieee.org/abstract/document/10645586
• L. Li*, E. Piccoli*, A. Cossu, D. Bacciu, V. Lomonaco. Calibration of Continual Learning Models. CLVision workshop, CVPR 2024. (* Equal contribution.)
  • Provides an empirical study of calibration in continual learning and proposes a continual calibration approach to make model confidence more reliable over time.
  • Paper link: https://ieeexplore.ieee.org/document/10678294