@inproceedings{raptors2026-c049,
  title     = {{LiDAR Challenges in Camera-Centric Perception}},
  author    = {Nagvekar, Anushree},
  booktitle = {Proceedings of the Raptors Conference},
  volume    = {1},
  year      = {2026},
  publisher = {Hackathon Raptors},
  address   = {London, United Kingdom},
  doi       = {10.68495/rc.2026.C049},
  url       = {https://doi.org/10.68495/rc.2026.C049},
  abstract  = {LiDAR provides essential 3D geometry for perception, but its role in camera-centric systems remains challenging due to sparse point clouds, limited semantic detail, weather sensitivity, and the cost of cross-sensor fusion. These issues can make tasks such as semantic segmentation and dense alignment more difficult, even though LiDAR still offers strong advantages for depth estimation, object separation, and tracking stability. The LiDAR reward: a 0.26 Gbps point cloud data replaces 3 Gbps of camera data plus 0.5–60 TFLOPS of depth reconstruction, with error flat across range. This talk covers selective fusion, adaptive attention, and compute-efficient processing to cut latency and bandwidth.}
}