Filippo Ghilotti

Filippo Ghilotti

I work on pseudo-labeling and holistic scene understanding for autonomous driving, building scalable training pipelines that turn raw multimodal data into reliable supervision. My goal is to break the labeling bottleneck, driving down cost while improving quality, through deep learning and classical optimization.

Contact

office
Room 1351 at ZIB
e-mail
languages
Italian and English

Curriculum vitae

since 2026
Researcher at ZIB

Publications and preprints

Preprints and reports

  1. Feng, B., Mei, Z., Ost, J., Ghilotti, F., Li, B., Girgis, R., Majumdar, A., and Heide, F. (2025). VERDI: VLM-Embedded Reasoning for Autonomous Driving. [arXiv]
    [BibTeX]
    @misc{2025_BowenEtAl_VerdiVlmreasoning_2505-15925,
      archiveprefix = {arXiv},
      eprint = {2505.15925},
      arxiv = {arXiv:2505.15925},
      primaryclass = {cs.RO},
      author = {Feng, Bowen and Mei, Zhiting and Ost, Julian and Ghilotti, Filippo and Li, Baiang and Girgis, Roger and Majumdar, Anirudha and Heide, Felix},
      title = {VERDI: VLM-Embedded Reasoning for Autonomous Driving},
      date = {2025-05-21}
    }
  2. Ghilotti, F., Palladin, E., Brucker, S., Sigal, A., Bijelic, M., and Heide, F. (2026). TruckDrive: Long-Range Autonomous Highway Driving Dataset. [arXiv]
    [BibTeX]
    @misc{2026_GhilottiEtAl_Truckdrivedataset_2603-02413,
      archiveprefix = {arXiv},
      eprint = {2603.02413},
      arxiv = {arXiv:2603.02413},
      primaryclass = {cs.CV},
      author = {Ghilotti, Filippo and Palladin, Edoardo and Brucker, Samuel and Sigal, Adam and Bijelic, Mario and Heide, Felix},
      title = {TruckDrive: Long-Range Autonomous Highway Driving Dataset},
      date = {2026-03-02}
    }

Conference contributions

  1. Ghilotti, F., Brucker, S., Saidy, N., Matteucci, M., Bijelic, M., and Heide, F. (2026). UniLiPs: Unified LiDAR Pseudo-Labeling with Geometry-Grounded Dynamic Scene Decomposition. 2026 International Conference on 3D Vision (3DV), 1680–1690. Associated with 3DV 2026. DOI: 10.1109/3dv69130.2026.00160 [arXiv]
    [BibTeX]
    @inproceedings{2026_GhilottiEtAl_Unilipspseudolabeling_2601-05105,
      year = {2026},
      booktitle = {2026 International Conference on 3D Vision (3DV)},
      publisher = {Institute of Electrical and Electronics Engineers},
      pages = {1680--1690},
      doi = {10.1109/3dv69130.2026.00160},
      archiveprefix = {arXiv},
      eprint = {2601.05105},
      arxiv = {arXiv:2601.05105},
      primaryclass = {cs.CV},
      author = {Ghilotti, Filippo and Brucker, Samuel and Saidy, Nahku and Matteucci, Matteo and Bijelic, Mario and Heide, Felix},
      title = {UniLiPs: Unified LiDAR Pseudo-Labeling with Geometry-Grounded Dynamic Scene Decomposition}
    }
  2. Palladin, E., Brucker, S., Ghilotti, F., Narayanan, P., Bijelic, M., and Heide, F. (2025). Self-Supervised Sparse Sensor Fusion for Long Range Perception. 2025 IEEE/CVF International Conference on Computer Vision (ICCV), 27498–27509. Associated with ICCV 2025. DOI: 10.1109/iccv51701.2025.02553 [arXiv]
    [BibTeX]
    @inproceedings{2025_PalladinEtAl_Selfsupervisedfusion_2508-13995,
      year = {2025},
      booktitle = {2025 IEEE/CVF International Conference on Computer Vision (ICCV)},
      publisher = {Institute of Electrical and Electronics Engineers},
      pages = {27498--27509},
      doi = {10.1109/iccv51701.2025.02553},
      archiveprefix = {arXiv},
      eprint = {2508.13995},
      arxiv = {arXiv:2508.13995},
      primaryclass = {cs.CV},
      author = {Palladin, Edoardo and Brucker, Samuel and Ghilotti, Filippo and Narayanan, Praveen and Bijelic, Mario and Heide, Felix},
      title = {Self-Supervised Sparse Sensor Fusion for Long Range Perception}
    }

Projects

GENAD
Generative AI for Scalable Autonomous Driving Systems

The project aims at shaping the next generation of highway autonomy for heavy trucks by combining long-range multimodal data, controllable generative models, a closed-loop virtual gym, and vision-language model (VLM) embedded reasoning. Rather than treating autonomy as a pure perception or planning problem, the project aims to advance generative simulation, reinforcement learning and chain-of-thought reasoning and language-guided representation learning as a unified research agenda.

GENAD
Mar 2026 to Mar 2029
5