About me

Hi, I’m Lucas Nunes! I’m a Ph.D. student at the University of Bonn in Germany, supervised by Prof. Dr. Cyrill Stachniss. I completed my Master’s at the University of São Paulo in Brazil, where I was born. Besides being a researcher, I’m also a photography enthusiast and enjoy playing video games.

My research interest lies in Generative Models and Representation Learning for large-scale 3D data in the autonomous driving context (but not only).

• Representation Learning: My research in the field is related to unsupervised representation learning from unlabeled data in the context of autonomous driving.
  • SegContrast: SegContrast: 3D Point Cloud Feature Representation Learning through Self-supervised Segment Discrimination
  • TARL: Temporal Consistent 3D LiDAR Representation Learning for Semantic Perception in Autonomous Driving
  • 3DUIS: Unsupervised Class-Agnostic Instance Segmentation of 3D LiDAR Data for Autonomous Vehicles
• Generative Models: I have started working with generative models during my master’s. Currently, my research in the field is in the area of 3D semantic scene-scale data generation.
  • LiDiff: Scaling Diffusion Models to Real-World 3D LiDAR Scene Completion
  • 3DiSS: Towards Generating Realistic 3D Semantic Training Data for Autonomous Driving
• Besides the main projects that I have worked on during my Ph.D. listed above, I have also collaborated on other projects with open-sourced code listed below:
  • ContMAV: Open-World Semantic Segmentation Including Class Similarity
  • Mask4D: End-to-End Mask-Based 4D Panoptic Segmentation for LiDAR Sequences
  • MaskPLS: Mask-Based Panoptic LiDAR Segmentation for Autonomous Driving
  • KPPR: Exploiting Momentum Contrast for Point Cloud-Based Place Recognition
  • 4DMOS: Receding Moving Object Segmentation in 3D LiDAR Data Using Sparse 4D Convolutions
  • Auto-MOS: Automatic Labeling to Generate Training Data for Online LiDAR-based Moving Object Segmentation

Link to my CV.