Benjamin Caine
Papers
1
Total Citations
71
H-Index
1
About
Benjamin Caine is a leading researcher in 3D perception for autonomous systems, whose work centers on efficient, real-time object detection from LiDAR and range sensor data. His most influential contribution, the 2021 paper *"To the Point: Efficient 3D Object Detection in the Range Image with Graph Convolution Kernels"* (71 citations), introduced a novel paradigm: rather than converting range images into dense 3D voxel grids or point clouds, Caine’s architecture directly learns 3D representations from the native 2D range image view. By embedding spherical coordinates into a 2D convolutional network, his method dramatically reduces computational overhead while preserving geometric fidelity—a breakthrough that enables faster, more accurate detection for resource-constrained robotics platforms. This work has become a cornerstone for researchers seeking to bridge the gap between sensor-native data formats and deep learning. Caine’s broader impact is evident in the adoption of his approaches across autonomous driving and mobile robotics, where efficiency is paramount. His research continues to shape how machines perceive depth and structure from sparse, real-world sensor streams, making him a pivotal figure in the push toward deployable, high-performance 3D vision systems.
Research Focus
Key Achievements
Top Papers
- 1